Madhumangal Pal
Department of Applied Mathematics with
Oceanology and Computer Programming
Vidyasagar University
Midnapore  721102
Dedicated to my parents
Preface
Numerical Analysis is a multidisciplinary subject. It has formed an integral part of
the undergraduate and postgraduate curriculum in Mathematics, Computer Science,
Physics, Commerce and dierent Engineering streams. Numerical Analysis shows the
way to obtain numerical answers to applied problems. Numerical methods stand there
where analytical methods may fail or complicated to solve the problem. For example, in
nding the roots of transcendental equations or in solving nonlinear dierential equations. So, it is quite impossible to train the students in applied sciences or engineering
without an adequate knowledge of numerical methods.
The book is suitable for undergraduate as well as for postgraduate students and
advanced readers. Ample material is presented so that instructors will able to select
topics appropriate to their needs. The book contains ten chapters.
In Chapter 1, dierent types of errors and their sources in numerical computation
are presented. The representation of oating point numbers and their arithmetic are
studied in this chapter.
Finite dierence operators, relations among them are well studied in Chapter 2. The
dierence equations and their solution methods are also introduced here.
Chapter 3 is devoted to single and bivariate interpolations. Dierent types of interpolation methods such as Lagrange, Newton, Bessal, Stirling, Hermite, Everette are
incorporated here. Inverse and cubic spline interpolation techniques are also presented
in this chapter. Several bivariate methods are presented here.
Several methods such as graphical, tabulation, bisection, regulafalsi, xed point
iteration, NewtonRaphson, Aitken, secant, Chebyshev and Muller are well studied to
solve an algebraic and transcendental equation in Chapter 4. The geometrical meaning
and the rate of convergence of the above methods are also presented. The very new
method, modied NewtonRaphson with cubic convergence is introduced here. BirgeVieta, Bairstow and Graees root squaring methods are deduced and illustrated to
nd the roots of a polynomial equation. The methods to solve a system of nonlinear
equations are introduced here.
Chapter 5 deals to solve a system of linear equations. Dierent direct and iterative
methods such as matrix inverse, GaussJordon, Gauss elimination, LU decomposition,
vii
ix
I feel great reverence for my parents, sisters, sisterinlaw and relatives for their
blessings and being a constant source of inspiration.
I would like to thank to Sk. Md. Abu Nayeem, Dr. Amiya K. Shyamal, Dr. Anita
Saha, for scrutinizing the manuscript.
I shall feel great to receive constructive criticisms for the improvement of the book
from the experts as well as the learners.
I thank the Narosa Publishing House Pvt. Ltd. for their sincere care in the
publication of the book.
Madhumangal Pal
x Numerical Analysis
Contents
1 Errors in Numerical Computations
1.1 Sources of Errors . . . . . . . . . . . . . . . . . . . . . . . .
1.2 Exact and Approximate Numbers . . . . . . . . . . . . . . .
1.3 Absolute, Relative and Percentage Errors . . . . . . . . . .
1.4 Valid Signicant Digits . . . . . . . . . . . . . . . . . . . . .
1.5 Propagation of Errors in Arithmetic Operations . . . . . . .
1.5.1 The errors in sum and dierence . . . . . . . . . . .
1.5.2 The error in product . . . . . . . . . . . . . . . . . .
1.5.3 The error in quotient . . . . . . . . . . . . . . . . . .
1.5.4 The errors in power and in root . . . . . . . . . . . .
1.5.5 Error in evaluation of a function of several variables
1.6 Signicant Error . . . . . . . . . . . . . . . . . . . . . . . .
1.7 Representation of Numbers in Computer . . . . . . . . . . .
1.8 Arithmetic of Normalized Floating Point Numbers . . . . .
1.8.1 Addition . . . . . . . . . . . . . . . . . . . . . . . . .
1.8.2 Subtraction . . . . . . . . . . . . . . . . . . . . . . .
1.8.3 Multiplication . . . . . . . . . . . . . . . . . . . . . .
1.8.4 Division . . . . . . . . . . . . . . . . . . . . . . . . .
1.9 Eect of Normalized Floating Point Representations . . . .
1.9.1 Zeros in oating point numbers . . . . . . . . . . . .
1.10 Exercise . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2 Calculus of Finite Di. and Di. Equs
2.1 Finite Dierence Operators . . . . . . . . . . .
2.1.1 Forward dierences . . . . . . . . . . . .
2.1.2 Backward dierences . . . . . . . . . . .
2.1.3 Central dierences . . . . . . . . . . . .
2.1.4 Shift, Average and Dierential operators
2.1.5 Factorial notation . . . . . . . . . . . .
2.2 Properties of Forward Dierences . . . . . . . .
xi
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
1
1
2
4
7
9
9
10
11
14
15
16
18
19
19
20
21
21
22
23
23
.
.
.
.
.
.
.
27
27
27
28
29
30
31
32
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
33
34
39
43
45
47
52
53
55
55
56
63
66
3 Interpolation
3.1 Lagranges Interpolation Polynomial . . . . . . . . . . . . . . . . .
3.1.1 Lagrangian interpolation formula for equally spaced points
3.2 Properties of Lagrangian Functions . . . . . . . . . . . . . . . . . .
3.3 Error in Interpolating Polynomial . . . . . . . . . . . . . . . . . . .
3.4 Finite Dierences . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.4.1 Forward dierences . . . . . . . . . . . . . . . . . . . . . . .
3.4.2 Backward dierences . . . . . . . . . . . . . . . . . . . . . .
3.4.3 Error propagation in a dierence table . . . . . . . . . . . .
3.5 Newtons Forward Dierence Interpolation Formula . . . . . . . .
3.5.1 Error in Newtons forward formula . . . . . . . . . . . . . .
3.6 Newtons Backward Dierence Interpolation Formula . . . . . . . .
3.6.1 Error in Newtons backward interpolation formula . . . . .
3.7 Gaussian Interpolation Formulae . . . . . . . . . . . . . . . . . . .
3.7.1 Gausss forward dierence formula . . . . . . . . . . . . . .
3.7.2 Remainder in Gausss forward central dierence formula . .
3.7.3 Gausss backward dierence formula . . . . . . . . . . . . .
3.7.4 Remainder of Gausss backward central dierence formula .
3.8 Stirlings Interpolation Formula . . . . . . . . . . . . . . . . . . . .
3.9 Bessels Interpolation Formula . . . . . . . . . . . . . . . . . . . . .
3.10 Everetts Interpolation Formula . . . . . . . . . . . . . . . . . . . .
3.10.1 Relation between Bessels and Everetts formulae . . . . . .
3.11 Interpolation by Iteration (Aitkens Interpolation) . . . . . . . . .
3.12 Divided Dierences and their Properties . . . . . . . . . . . . . . .
3.12.1 Properties of divided dierences . . . . . . . . . . . . . . .
3.13 Newtons Fundamental Interpolation Formula . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
71
72
75
77
79
87
87
88
88
90
92
95
98
99
99
102
102
105
105
106
109
111
113
116
117
121
Contents
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
xiii
.
.
.
.
.
.
.
.
124
124
124
125
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
126
131
132
132
134
135
136
139
141
142
156
156
159
167
181
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
189
190
190
191
193
198
202
204
208
210
213
218
219
222
226
226
230
230
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
237
238
239
245
253
261
261
262
266
271
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
275
276
277
278
287
287
293
297
302
304
304
313
314
317
320
325
326
327
329
330
331
334
338
339
344
346
352
354
358
Contents
xv
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
365
365
368
372
374
374
375
380
380
381
390
392
394
401
403
. . . . . . 403
. . . . . . 403
. . . . . . 404
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
407
410
413
419
420
424
430
431
432
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
433
433
438
445
446
446
448
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
449
450
452
453
455
456
462
464
466
468
469
470
473
480
486
486
490
492
495
497
502
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
511
513
515
517
518
520
522
526
526
528
533
537
538
539
540
541
541
547
8.7
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Contents
xvii
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
552
553
555
560
563
571
572
572
572
573
575
577
577
.
.
.
.
.
.
.
.
.
.
583
585
586
586
588
597
599
600
605
615
617
.
.
.
.
.
.
.
.
.
.
.
.
.
619
619
620
623
625
626
626
627
628
628
628
630
633
636
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
637
640
645
646
Description
Lagranges interpolation for single variable
Newtons forward interpolation
Aitkens interpolation
Interpolation by divided dierence
Interpolation by cubic spline
Lagrange bivariate interpolation
Solution of an equation by bisection method
Solution of an equation by RegulaFalsi method
Solution of an equation by xed point iteration method
Solution of an equation by NewtonRaphson method
Solution of an equation by secant method
Roots of polynomial equation by BirgeVirta method
Roots of polynomial equation by Bairstow method
Seidal iteration method for a pair of nonlinear equations
NewtonRapshon method for a pair of equations
Determinant using partial pivoting
Determinant using complete pivoting
Determination of matrix inverse
Solution of a system of equations by matrix inverse
method
Solution of a system of equations by Gauss elimination
method
Solution of a system of equations by LU decomposition
method
Solution of a tridiagonal system of equations
Solution of a system of equations by GaussJacobis iteration
Solution of a system of equations by GaussSeidals iteration
xix
Algo.
85
93
115
129
149
162
197
200
207
224
228
243
250
264
268
282
284
290
294
Prog.
85
94
115
130
151
163
197
201
208
225
229
244
251
264
269
283
285
291
294
300
301
310
311
323
341
323
342
350
351
xx Numerical Analysis
Sl. No.
5.20
6.1
6.2
6.3
6.4
7.1
7.2
7.3
7.4
7.5
7.6
7.7
7.8
8.1
8.2
8.3
8.4
8.5
8.6
8.7
9.1
9.2
10.1
10.2
Description
Solution of a system of equations by GaussSeidal SOR
method
Characteristic polynomial of a matrix by LeverrierFaddeev method
Largest eigenvalue by power method
Eigenvalue of a real symmetric matrix by Jacobis method
Eigenvalue of a real symmetric matrix by Householder
method
First derivative based on Lagranges interpolation
First derivative using Richardson extrapolation
Integration by trapezoidal rule
Integration by Simpsons 1/3 rule
GaussLegendre quadrature
Rombergs integration
Double integration using trapezoidal rule
Integration by Monte Carlo method
Solution of a rst order dierential equation by Eulers
method
Solution of a rst order dierential equation by modied
Eulers method
Solution of a rst order dierential equation by fourth
order RungeKutta method
Solution of a pair of rst order dierential equation by
RungeKutta method
Solution of a rst order dierential equation by AdamsBashforthMoulton method
Solution of a rst order dierential equation Milnes
predictorcorrector method
Solution of a second order BVP using nite dierence
method
Solution of heat equation using CrankNicolson implicit
method
Solution of Poissons equation using GaussSeidal S.O.R.
method
Fitting of straight line by least square method
Approximation of a function by Chebyshev polynomial
Algo.
Prog.
357
370
370
378
386
398
379
387
399
415
428
437
444
461
483
488
494
519
416
429
438
444
461
484
489
496
519
524
524
531
532
535
536
544
545
550
550
557
557
593
594
612
612
622
642
623
643
Chapter 1
Errors in Numerical
Computations
The solutions of mathematical problems are of two types: analytical and numerical.
The analytical solutions can be expressed in closed form and these solutions are error
free. On the other hand, numerical method is a division of mathematics which solves
problems using computational machine (computer, calculator, etc.). But, for some
classes of problems it is very dicult to obtain an analytical solution. For example, the
Indian populations are known at the years 1951, 1961, 1971, 1981, 1991, 2001. There
is no analytical method available to determine the population in the year, say, 2000.
But, using numerical method one can determine the population in the said year. Again,
sometimes we observed that the solutions of nonlinear dierential equations cannot be
determined by analytical methods, but, such problems can easily be solved by numerical
methods. Numerical computations are almost invariably contaminated by errors, and
it is important to understand the source, propagation, magnitude, and rate of growth
of these errors.
In this age of computer, many complicated and large problems are solved in signicantly less time. But, without using numerical methods we cannot solve any mathematical problem using computer, as analytical methods are not suitable to solve a problem
by computer. Thus, the numerical methods are highly appreciated and extensively used
by Mathematicians, Computer Scientists, Statisticians, Engineers and others.
1.1
Sources of Errors
2 Numerical Analysis
and their growth and propagation in numerical computation. Three types of errors, viz.,
inherent errors, roundo errors and truncation errors, occur in nding the solution of
a problem using numerical method. These three type of errors are discussed below.
(i) Inherent errors: This type of errors is present in the statement of the problem itself,
before determining its solution. Inherent errors occur due to the simplied assumptions
made in the process of mathematical modelling of a problem. It can also arise when the
data is obtained from certain physical measurements of the parameters of the proposed
problem.
(ii) Roundo errors: Generally, the numerical methods are carried out using calculator or computer. In numerical computation, all the numbers are represented by
decimal fraction. Some numbers such as 1/3, 2/3, 1/7 etc. can not be represented by
decimal fraction in nite numbers of digits. Thus, to get the result, the numbers should
be roundedo into some nite number of digits.
Again, most of the numerical computations are carried out using calculator and computer. These machines can store the numbers up to some nite number of digits. So in
arithmetic computation, some errors will occur due to the nite representation of the
numbers; these errors are called roundo error. Thus, roundo errors occur due to the
nite representation of numbers during arithmetic computation. These errors depend
on the word length of the computational machine.
(iii) Truncation errors: These errors occur due to the nite representation of an
inherently innite process. For example, the use of a nite number of terms in the
innite series to compute the value of cos x, sin x, ex , etc.
The Taylors series expansion of sin x is
sin x = x
x3 x5 x7
+
+ .
3!
5!
7!
This is an innite series expansion. If only rst ve terms are taken to compute the
value of sin x for a given x, then we obtain an approximate result. Here, the error occurs
due to the truncation of the series. Suppose, we retain the rst n terms, the truncation
error (Etrunc ) is given by
Etrunc
x2n+1
.
(2n + 1)!
It may be noted that the truncation error is independent of the computational machine.
1.2
To solve a problem, two types of numbers are used. They are exact and approximate.
Exact number gives a true value of a result and approximate number gives a value which
is closed to the true value.
For example, in the statements a triangle has three sides, there are 2000 people in a
locality, a book has 450 pages the numbers 3, 2000 and 450 are exact numbers. But,
in the assertions the height of a pupil is 178 cm, the radius of the Earth is 6400 km,
the mass of a match box is ten gram, the numbers 178, 6400 and 10 are approximate
numbers.
This is due to the imperfection of measuring instruments we use. There are no
absolutely exact measuring instruments; each of them has its own accuracy. Thus, the
height of a pupil is 178 cm is not absolute measurement. In the second example, the
radius of the Earth is very concept; actually, the Earth is not a sphere at all, and we
can use its radius only in approximate terms. In the last example, the approximation
of the number is also dened by the fact that dierent boxes may have dierent masses
and the number 10 denes the mass of a particular box.
One important observation is that, same number may be exact as well as approximate.
For example, the number 3 is exact when it represents the number of sides of a triangle
and approximate if we use it to represent the number when calculating the area of a
circle using the formula r2 .
4 Numerical Analysis
This rule is often called a rule of an even digit. If a number is rounded using the
above rule then the number is called correct up to some (say n) signicant gures.
The following numbers are roundedo correctly to ve signicant gures:
Exact number
25.367835
28.353215
3.785353
5.835453
6.73545
4.83275
0.005834578
3856754
2.37
8.99997
9.99998
Roundo number
25.368
28.353
3.7854
5.8355
6.7354
4.8328
0.0058346
38568102
2.3700
9.0000
10.000
From above examples, it is easy to observe that, while rounding a number, an error
is generated and this error is sometimes called roundo error.
1.3
Let xT be the exact value of a number and xA be its approximate value. If xA < xT ,
then we say that the number xA is an approximate value of the number xT by defect
and if xA > xT , then it is an approximate value of xT by excess.
The dierence between the exact value xT and its approximate value xA is an error.
As a rule, it is not possible to determine the value of the error xT xA and even its
sign, since the exact number xT is unknown.
The errors are represented in three ways, viz., absolute error, relative error and
percentage error.
Absolute error:
The absolute error of the approximate number xA is a quantity (x) which satises the
inequality
x xT xA .
The absolute error is the upper bound of the deviation of the exact number xT from
its approximation, i.e.,
xA x xT xA + x.
The above result can be written in the form
xT = xA x.
(1.1)
In other words, the absolute error of the number x is the dierence between true
value and approximate value, i.e.,
x = xT xA .
It may be noted from the rounding process that, if a number be rounded to m decimal
places then
1
absolute error 10m .
(1.2)
2
The absolute error measures only the quantitative aspect of the error but not the
qualitative one, i.e., does not show whether the measurement and calculation were
accurate. For example, the length and the width of a table are measured with a scale
(whose division is 1 cm) and the following results are obtained: the width w = 5 0.5
cm and the length l = 100 0.5 cm. In both cases the absolute error is same and it is
0.5 cm. It is obvious that the second measurement was more accurate than the rst.
To estimate the quality of calculations or measurements, the concept of a relative error
is introduced.
Relative error:
The relative error (x) of the number xA is
x =
x
x
or
, xT  = 0 and xA  = 0.
xA 
xT 
6 Numerical Analysis
Note 1.3.1 The absolute error of a number correct to n signicant gures cannot be
greater than half a unit in the nth place.
Note 1.3.2 The relative error and percentage error are independent of the unit of
measurement, while absolute error depends on the measuring unit.
1 0.999
1
0.333 =
3
3
0.001
= 0.00033.
3
x
0.00033
= 0.00099 0.001.
=
xT
1/3
Example 1.3.3 Determine the absolute error and the exact number corresponding
to the approximate number xA = 5.373 if percentage error is 0.01%.
Solution. Here the relative error x = 0.01% = 0.0001.
The absolute error x = xA x = 5.373 0.0001 = 0.0005373 0.00054.
The exact value = 5.373 0.00054.
Example 1.3.4 Find out in which of the following cases, the quality of calculations
15
0.8824 and yT = 51 7.141.
is better: xT =
17
Solution. To nd the absolute error, we take thenumbers xA and yA with a larger
number of decimal digits as xA 0.882353, yA = 51 7.141428.
Therefore, the absolute error in xT is 0.882353 0.8824 0.000047,
and in yT is 7.141428 7.141 0.00043.
The relative error in xA is 0.00047/0.8824 0.00053 = 0.05%
and relative error in yA is 0.00043/7.141 = 0.0000602 = 0.006%.
In the second case the quality of calculation is better than the rst case as relative
error in xT > relative error in yT .
1.4
A real number can be represented by many dierent ways. For example, the number
840000 can be represented as two factors: 840 103 or 84.0 104 or 0.840 106 . (Note
that in these representations the last three signicant zeros are lost). The later form of
the notation is known as normalize form and it is commonly used. In this case, we
say that 840 is the mantissa of the number and 6 is its order.
Every positive decimal number, exact as well as approximate, can be expressed as
a = d1 10m + d2 10m1 + + dn 10mn+1 + ,
where di are the digits constituting the number (i = 1, 2, . . .) with d1 = 0 and 10mi+1
is the value of the ith decimal position (counting from left).
The digit dn of the approximate number a is valid signicant digit (or simply a
valid digit) if it satises the following inequality.
a 0.5 10mn+1 ,
(1.3)
i.e., absolute error does not exceed half the unit of the decimal digit in which dn appears.
If inequality (1.3) is not satised, then the digit dn is said to be doubtful. It is
obvious that if the digit dn is valid, then all the preceding digits, to the left of it, are
also valid.
8 Numerical Analysis
Theorem 1.1 If a number is correct up to n signicant gures and the rst signicant
digit of the number is k, then the relative error is less than
1
.
k 10n1
Proof. Let xA be the approximate value of the exact number xT . Also, let xA is correct
up to n signicant gures and m decimal places. Then there are three possibilities may
occur:
(i) m < n
(ii) m = n and
(iii) m > n.
We have by (1.2), the absolute error x 0.5 10m .
Case I. When m < n.
In this case, the total number of digits in integral part is n m. If k be the rst
signicant digit in xT , then
x 0.5 10m and xT  k 10nm1 0.5 10m .
Therefore, the relative error
x =
0.5 10m
x
xT 
k 10nm1 0.5 10m
1
.
=
2k 10n1 1
1
.
k 10n1
1.5
1.5.1
(1.4)
Thus the absolute error in sum of approximate numbers is equal to the sum of the
absolute errors of the numbers.
From (1.4), it follows that the absolute error of the algebraic sum must not be smaller
than the absolute error of the least exact term.
The following points should be kept in mind when adding numbers of dierent absolute accuracy.
(i) identify a number (or numbers) of the least accuracy (i.e., a number which has
the maximum absolute error),
(ii) roundo more exact numbers so as to retain in them one digit more than in the
identied number (i.e., retain one reserve digit),
(iii) perform addition taking into account all the retained digits,
(iv) roundo the result by discarding one digit.
10 Numerical Analysis
Subtraction
Consider x1 and x2 be two approximate values of the corresponding exact numbers X1
and X2 . Let X = X1 X2 and x = x1 x2 .
Then X1 = x1 x1 and X2 = x2 x2 , where x1 and x2 are the errors in x1
and x2 respectively.
Therefore, X x = (X1 x1 ) (X2 x2 ) X1 x1  + X2 x2 . Hence,
x = x1  + x2 .
(1.5)
Thus the absolute error in dierence of two numbers is equal to the sum of individual
absolute errors.
1.5.2
Let us consider two exact numbers X1 and X2 and their approximate values x1 and x2 .
Also, let x1 and x2 be the errors in x1 and x2 , i.e., X1 = x1 x1 and X2 = x2 x2 .
Now, X1 X2 = x1 x2 x1 x2 x2 x1 x1 x2 .
Then X1 X2 x1 x2  x1 x2  + x2 x1  + x1 x2 . The last term of right hand
side is small, so we discard it and dividing both sides by x = x1 x2 .
Thus the relative error in the product is
X1 X2 x1 x2 x2 x1
=
(1.6)
x2 + x1 .
x1 x2
Thus the relative errors in product of two numbers is equal to the sum of individual
relative errors.
The result (1.6) can be easily extended to the product of several numbers so that, if
X = X1 X2 Xn and x = x1 x2 xn , then
X x x1 x2
+
+ + xn .
=
(1.7)
x x1 x2
xn
That is, the total relative error in product of n numbers is equal to the sum of
individual relative errors.
A particular case
Let the approximate numbers x1 , x2 , . . . , xn be all positive and x = x1 x2 xn .
Then log x = log x1 + log x2 + + log xn .
x
x1 x2
xn
=
+
+ +
.
x
x1
x2
xn
x x x
x
1
2
n
That is,
=
+
+ +
.
x
x1
x2
xn
Usually, the following steps are followed when multiplying two numbers:
Therefore,
11
Let us consider two exact numbers X1 and X2 and their approximate values x1 and x2 .
x1
X1
and x =
.
Also, let X =
X2
x2
Then X1 = x1 + x1 , X2 = x2 + x2 , where x1 and x2 are the errors.
Let x1 = 0 and x2 = 0.
Now,
x2 x1 x1 x2
x1 + x1 x1
.
=
X x=
x2 + x2 x2
x2 (x2 + x2 )
Dividing both sides by x and taking absolute values:
X x x2 x1 x1 x2
x1 x2
x2
=
=
x x1 (x2 + x2 ) x2 + x2 x1 x2 .
The error x2 is small as compared to x2 , then approximately
x2
1. Therefore, above relation becomes
x2 + x2
x X x x1 x2 x1 x2
,
x =
+
=
=
x x x1
x2 x1 x2
(1.8)
i.e., x = x1 + x2 . Hence, the total relative error in quotient is equal to the sum of
their individual relative errors.
12 Numerical Analysis
The relation (1.8) can also be written as
x x1 x2 x1 x2
=
x x1 x2 x1 x2 .
(1.9)
From this relation one can conclude that the relative error in quotient is greater than
or equal to the dierence of their individual relative errors.
A particular case
For positive approximate numbers x1 and x2 , the equation (1.8) can easily be deduced.
Let x = x1 /x2 . Then log x = log x1 log x2 . Therefore,
x x1 x2
x1 x2
x
.
=
i.e.,
+
x
x1
x2
x x1 x2
While dividing two numbers the following points should be followed.
(i) identify the least exact number, i.e., the number with the least number of valid
digits,
(ii) roundo the other number, leaving in it on signicant digit more than there are
digits in the identied number,
(iii) retain as many signicant digits in the quotient as there were in the least exact
number.
Example 1.5.2 Find the sum of the approximate numbers 0.543, 0.1834, 17.45,
0.000234, 205.2, 8.35, 185.3, 0.0863, 0.684, 0.0881 in each of which all the written
digits are valid. Find the absolute error in sum.
Solution. The least exact numbers (those possessing the maximum absolute error)
are 205.2 and 185.3. The error of each of them is 0.05. Now, rounding o the other
numbers, leaving one digit more and adding all of them.
0.54+0.18+17.45+0.00+205.2+8.35+185.3+0.09+0.68+0.09=417.88.
Discarding one digit by roundo the sum and we obtained 417.9.
The absolute error in the sum consists of two terms:
(i) the initial error, i.e., the sum of the errors of the least exact numbers and the
rounding errors of the other numbers: 0.05 2 + 0.0005 8 = 0.104 0.10.
(ii) the error in roundingo the sum is 417.9 417.88 = 0.02.
Thus the absolute error of the sum is 0.10 + 0.02 = 0.12.
So, the sum can be written as 417.9 0.12.
13
Example 1.5.3 Find the dierence of the approximate numbers 27.5 and 35.8 having absolute errors 0.02 and 0.03 respectively. Evaluate the absolute and the relative
errors of the result.
Solution. Let x1 = 27.5 and x2 = 35.8. Then x = x1 x2 = 8.3. The total
absolute error x = 0.02 + 0.03 = 0.05.
Thus the dierence x1 x2 is 8.3 with absolute error 0.05.
The relative error is 0.05/ 8.3 0.006 = 0.6%.
Example 1.5.4 Find the product of the approximate numbers x1 = 8.6 and x2 =
34.359 all of whose digits are valid. Also nd the relative and the absolute errors.
Solution. In the rst number, there are two valid signicant digits and in the second
there are ve digits. Therefore, roundo the second number to three signicant
digits. After roundingo the numbers x1 and x2 become x1 = 8.6 and x2 = 34.4.
Hence the product is
x = x1 x2 = 8.6 34.4 = 295.84 3.0 102 .
In the result, there are two signicant digits, because the least number of valid signicant digits of the given numbers is 2.
The relative error in product is
x x1 x2 0.05 0.0005
=
+
= 0.00583 0.58%.
+
=
x =
x x1 x2
8.6
34.359
The absolute error is (3.0 102 ) 0.00583 = 1.749 1.7.
Example 1.5.5 Calculate the quotient x/y of the approximate numbers x = 6.845
and y = 2.53 if all the digits of the numbers are valid. Find the relative and the
absolute errors.
Solution. Here the dividend x = 6.845 has four valid signicant digits and the
divisor has three, so we perform division without roundingo. Thus
6.845
x
=
= 2.71.
y
2.53
Three signicant digits are retained in the result, since, the least exact number (the
divisor y) contains three valid signicant digits.
The absolute error in x and y are respectively
x = 0.0005 and y = 0.005.
14 Numerical Analysis
Therefore the relative error in quotient is
x y 0.0005 0.005
+
x y = 6.845 + 2.53 = 0.000073 + 0.00198
0.002 = 0.2%.
The absolute error is
x
0.002 = 2.71 0.002 = 0.00542 = 0.005.
y
1.5.4
Let us consider an approximate number x1 which has a relative error x1 . Now, the
problem is to nd the relative error of x = xm
1 .
Then
x1 x1 x1 .
x = xm
1 =
m times
(1.10)
m times
Thus, when the approximate number x is raised to the power m, its relative error
increases m times.
x
x
1 x1
1 x1
or
=
.
=
x
m x1
x m x1
1
x1 ,
m
15
X3 Y
Example 1.5.6 Calculate A =
where X = 8.36, Y = 80.46, Z = 25.8. The
Z2
absolute errors in X, Y, Z are respectively 0.01, 0.02 and 0.03. Find the error of the
result.
Solution. Here the absolute error x = 0.01, y = 0.02 and z = 0.03. To calculate
intermediate result, retain one reserve digit. The approximate intermediate values
584.3 8.9699
= 7.87.
665.6
Three signicant digits are taken in the result, since, the least number of signicant
digits in the numbers is 3.
Now, the relative error a in a is given by
0.01 1
0.02
0.03
1
y + 2 z = 3
+
+2
2
8.36 2 80.46
25.8
0.0036 + 0.00012 + 0.0023 0.006 = 0.6%.
a = 3 x +
16 Numerical Analysis
This formula gives the total error for computing a function containing several variables.
The relative error is given by
y f xi
=
.
y
xi y
n
i=1
1.6
Signicant Error
Signicant error occurs due to the loss of signicant digits during arithmetic computation. This error occurs mainly due to the nite representation of the numbers in
computational machine (computer or calculator). The loss of signicant digits occurs
due to the following two reasons:
(i) when two nearly equal numbers are subtracted and
(ii) when division is made by a very small divisor compared to the dividend.
Signicant error is more serious than roundo error, which are illustrated in the
following examples:
Example 1.6.1 Find the dierence X = 5.36 5.35 and evaluate the relative
error of the result.
X = 5.36 5.35 =
5.36 + 5.35
0.01
5.36 5.35
=
0.002 = x (say).
=
5.36 + 5.35
5.36 + 5.35
In this case the relative error is
x =
0.001
x1 + x2
= 0.0002 = 0.02%.
=
x1 + x2
2.315 + 2.313
Thus, when calculating x1 and x2 with the same four digits we get a better result in
the sense of a relative error.
17
Example 1.6.2 Calculate the values of the function y = 1 cos x at x = 82o and
at x = 1o . Also, calculate the absolute and the relative errors of the results.
Solution. y at x = 82o
The value of cos 82o 0.1392 = a1 (say) (correct up to four digits) and a1 =
0.00005. Then y1 = 1 0.1392 = 0.8608 and y1 = 0.00005 (from an exact number equal to unity we subtract an approximate number with an absolute error not
exceeding 0.00005).
Consequently, the relative error is
y1 =
0.00005
= 0.000058 = 0.006%.
0.8608
y at x = 1o
We have cos 1o 0.9998 = a2 (say). a2 = 0.00005.
y2 = 1 0.9998 = 0.0002. y2 = 0.00005.
Hence
0.00005
= 0.25 = 25%.
y2 =
0.0002
From this example it is observed that for small values of x, a direct calculation of
y = 1 cos x gives a relative error of the order 25%. But at x = 82o the relative error
is only 0.006%.
Now, change the calculation procedure and use the formula y = 1 cos x = 2 sin2 x2
to calculate the value of y for small values of x.
Let a = sin 0o 30 0.0087. Then a = 0.00005 and
a =
0.00005
= 0.0058 = 0.58%.
0.0087
1000 106 1
.
2
18 Numerical Analysis
Now,
106 1 = 0.1000 107 0.0000 107 = 0.1000 107 .
Thus 106 1 = 0.1000 104 .
0.1000 104 0.1000 104
Therefore the roots are
2
which are respectively 0.1000 104 and 0.0000 104 . One of the roots becomes zero
due to the nite representation of the numbers. But, the transformed formula gives
the smaller root more accurately.
The smaller root of the equation may be calculated using the transformed formula
=
2
2(1000 + 106 1)
1
=
= 0.00025.
2(1000 + 106 1)
Thus the roots of the given equation are 0.1000 104 and 0.00025.
Such a situation may be recognized by checking 4ac b2 .
It is not always possible to transform the computing formula. Therefore, when nearly
equal numbers are subtracted, they must be taken with a sucient number of reserve
valid digits. If it is known that the rst m signicant digits may be lost during computation and if we need a result with n valid signicant digits then the initial data should
be taken with m + n valid signicant digits.
1.7
19
oating point mode. In this mode of representation, the whole number is converted to
a proper fraction in such a way that the rst digit after decimal point should be nonzero
and is adjusted by multiplying a number which is some powers of 10. For example, the
number 375.3 104 is represented in this mode as .3753 107 = .3753E7 (E7 is used
to represent 107 ). From this example, it is observed that in normalized oating point
representation, a number is a combination of two parts mantissa and exponent.
In the above example, .3753 is the mantissa and 7 is the exponent. It may be noted
that the mantissa is always greater than or equal to .1 and exponent is an integer.
For simplicity, it is assume that the computer (hypothetical) uses four digits to store
mantissa and two digits for exponent. The mantissa and the exponent have their own
signs.
The number .0003783 would be stored as .3783E3. The leading zeros in this number
serve only to indicate the decimal point. Thus, in this notation the range of numbers
(magnitudes) is .9999 1099 to .1000 1099 .
1.8
In this section, the arithmetic operations on normalized oating point numbers are
discussed.
1.8.1
Addition
If two numbers have same exponent, then the mantissas are added directly and the
exponents are adjusted, if required.
If the exponents are dierent then lower exponent is shifted to higher exponent by
adjusting mantissa. The details about addition are discussed in the following examples.
Example 1.8.1 Add the following normalized oating point numbers.
(i) .3456E3 and .4325E3 (same exponent)
(ii) .8536E5 and .7381E5
(iii) .3758E5 and .7811E7 (dierent exponent)
(iv) .2538E2 and .3514E7
(v) .7356E99 and .3718E99 (overow condition)
Solution. (i) In this case, the exponents are equal, so the mantissa are added
directly. Thus the sum is .7781E3.
(ii) In this case, the exponent are equal and the sum is 1.5917E5. Here the mantissa
has 5 signicant digits, but our computer (hypothetical) can store only four signicant
gures. So, the number is shifted right one place before it is stored. The exponent is
increased by 1 and the last digit is truncated. The nal result is .1591E6.
20 Numerical Analysis
(iii) Here, the numbers are .3758E5 and .7811E7. The exponent of the rst number
is less than that of the second number. The dierence of the exponents is 7 5 = 2.
So the mantissa of the smaller number (here rst number) is shifted right by 2 places
(the dierence of the exponents) and the last 2 digits of the mantissa are discarded
as our hypothetical computer can store only 4 digits. Then the rst number becomes
.0037E7. Then the result is .0037E7 + .7811E7 = .7848E7.
(iv) Here also the exponents are dierent and the dierence is 7 2 = 5. The mantissa of rst number (smaller exponent) is shifted 5 places and the number becomes
.0000E7. The nal result is .0000E7 + .3514E7 = .3514E7.
(v) Here the numbers are .7356E99 and .3718E99 and they have equal exponent. So
the sum of them is 1.1074E99. In this case mantissa has ve signicant digits. Thus
the mantissa is shifted right and the exponent is increased by 1. Then the exponent
becomes 100. As the exponent cannot store more than two digits, in our hypothetical
computer, the number is larger than the largest number that can be stored in our
computer. This situation is called an overow condition and the machine will give
an error message.
1.8.2
Subtraction
The subtraction is same as addition. In subtraction one positive number and one negative number are added. The following example shows the details about subtraction.
Example 1.8.2 Subtract the normalized oating point numbers indicated below:
(i) .3628E6 from .8321E6
(ii) .3885E5 from .3892E5
(iii) .3253E7 from .4123E6
(iv) .5321E99 from .5382E99.
Solution. (i) Here the exponents are equal, and the result is
.8321E6 .3628E6 = .4693E6.
(ii) Here the result is .3892E5 .3885E5 = .0007E5. The most signicant digit in the
mantissa is 0, so the mantissa is shifted left till the most signicant digit becomes
nonzero and in each left shift of the mantissa the exponent is reduced by 1. Hence
the nal result is .7000E2.
(iii) The numbers are .4123E6 and .3253E7. The exponents are not equal, so the
number with smaller exponent is shifted right and the exponent increased by 1 for
every right shift. Then the second number becomes .0325E6. Thus the result is
.4123E6 .0325E6 = .3798E6.
21
(iv) The result is .5382E99 .5321E99 = .0061E99. For normalization, the mantissa is shifted left twice and in this process the exponent is reduced by 1. In rst
shift, the exponent becomes 100, but our hypothetical computer can store only two
digits as exponent. So 100 cannot be accommodated in the exponent part of the
number. In this case, the result is smaller than the smallest number which could be
stored in our computer. This condition is called an underow condition and the
computer will give an error message.
1.8.3
Multiplication
Two numbers in normalized oating point mode are multiplied by multiplying the mantissa and adding the exponents. After multiplication, the mantissa is converted into
normalized oating point form and the exponent is converted appropriately. The following example shows the steps of multiplication.
Example 1.8.3 Multiply the following numbers indicated below:
(i) .5321E5 by .4387E10
(ii) .1234E10 by .8374E10
(iii) .1139E50 by .8502E51
(iv) .3721E52 by .3205E53.
Solution. (i) Here, .5321E5 .4387E10 = .23343227
E15.
discarded
The mantissa has 8 signicant gures, so the last four digits are discarded. The nal
result is .2334E15.
(ii) Here, .1234E10 .8374E10 = .10333516
E0 = .1033E0.
discarded
Division
In the division, the mantissa of the numerator is divided by that of the denominator.
The exponent is obtained by subtracting exponent of denominator from the exponent
22 Numerical Analysis
of numerator. The quotient mantissa is converted to normalized form and the exponent
is adjusted appropriately.
Example 1.8.4 Perform the following divisions
(i) .9938E5 .3281E2
(ii) .9999E2 .1230E99
(iii) .3568E10 .3456E97.
Solution. (i) .9938E5 .3281E2 = .3028E4.
(ii) .9999E2 .1230E99 = .8129E102.
The result overows.
(iii) .3568E10 .3456E97 = .1032E106.
In this case the result underows.
1.9
It is very surprising that due to the truncation of mantissa, the associative and
distributive laws do not hold always in normalized oating point numbers.
That is,
(i) (a + b) + c = a + (b + c)
(ii) (a + b) c = (a c) + b
(iii) a(b c) = ab ac.
These results are illustrated in the following examples:
(i) a =.6878E1, b =.7898E1 and c =.1007E1.
Now, a + b =.1477E2
(a + b) + c = .1477E2 + .1007E1 = .1477E2 + .0100E2 = .1577E2.
Again, b + c =.8905E1.
a + (b + c)=.6878E1+.8905E1=.1578E2.
Thus, (a + b) + c = a + (b + c).
(ii) Let a =.6573E1, b =.5857E1, c =.6558E1.
Then a + b =.6631E1 and (a + b) c =.6631E1 .6558E1 = .7300E1.
Again, a c =.1500E1 and (a c) + b =.1500E1 + .5857E1 = .7357E1.
Thus, (a + b) c = (a c) + b.
(iii) Let a =.5673E1, b =.3583E1, c =.3572E1.
23
b c =.1100E1.
a(b c) =.5673E1 .1100E1 = .0624E0 = .6240E1.
ab =.2032E2, ac =.2026E2.
ab ac =.6000E1.
Thus, a(b c) = ab ac.
The above examples are intentionally chosen to point out the occurrence of inaccuracies in normalized oating point arithmetic due to the shifting and truncation of
numbers during arithmetic operations. But these situations always do not happen.
Here, we assume that the computer can store only four digits in mantissa, but actually
computer can store seven digits as mantissa (in single precision). The larger length of
mantissa gives more accurate result.
1.9.1
The number zero has a denite meaning in mathematics, but, in computer exact equality
of a number to zero can never be guaranteed. The cause behind this situation is that
most of the numbers in oating point representation are approximate. One interesting
example is presented below to illustrate the behaviour of zero.
1.10
Exercise
24 Numerical Analysis
3. Explain what do you understand by an approximate number and signicant gures
of a number.
4. What convention are used in roundingo a number ?
5. When a number is said to be correct to n signicant gures ?
Roundo the following numbers to three signicant gures.
(i) 0.01302, (ii) 349.87, (iii) 0.005922, (iv) 87678, (v) 64.8523, (vi) 6380.7, (vii)
0.0000098, (viii) .2345, (ix) 0.4575, (x) 34.653, (xi) 21.752, (xii) 1.99999.
6. Dene absolute, relative and percentage errors.
7. Explain when relative error is a better indicator of the accuracy of a computation
than the absolute error.
8. Find out which of the following two equalities is more exact:
(i)
6/25 1.4 or 1/3 0.333, (ii) 1/9 0.1 or 1/3 0.33, (iii) 3.142 or
10 3.1623.
9. Find the absolute, relative and percentage errors when (i) 2/3 is approximated
to 0.667, (ii) 1/3 is approximated to 0.333, and (iii) true value is 0.50 and its
calculated value was 0.49.
10. (i) If is approximated as 3.14 instead of 3.14156, nd the absolute, relative and
percentage errors.
(ii) Roundo the number x = 3.4516 to three signicant gures and nd the
absolute and the relative errors.
11. The numbers 23.982 and 3.4687 are both approximate and correct only to their
last digits. Find their dierence and state how many gures in the result are
trustworthy.
12. Two lengths X and Y are measured approximately up to three signicant gures
as X = 3.32 cm and Y = 5.39 cm. Estimate the error in the computed value of
X +Y.
13. Let xT and xA denote respectively the true and approximate values of a number.
Prove that the relative error in the product xA yA is approximately equal to the
sum of the relative errors in xA and yA .
14. Show that the relative error in the product of several approximate nonzero numbers does not exceed the sum of the relative errors of the numbers.
25
15. Show that the maximum relative error in the quotient of two approximate numbers
is approximately equal to the algebraic sum of the maximum relative errors of the
individual numbers.
16. Let x = 5.234 0.0005 and y = 5.123 0.0005. Find the percentage error of the
dierence a = x y when relative errors x = y = 0.0001.
17. What do you mean by the statement that xA (approximate value) has m signicant
gures with respect to xT (true value) ? If the rst signicant gure of xA is k
and xA is correct up to n signicant gures, prove that the relative error is less
than 101n /k.
18. Given a = 11 0.5, b = 0.04562 0.0001, c = 17200 100. Find the maximum
value of the absolute error in the following expressions
(i) a + 2b c, (ii) 2a 5b + c and (iii) a2 .
19. Calculate the quotient a = x/y of the approximate numbers x = 5.762 and y =
1.24 if all the digits of the dividend and the divisor are valid. Find the relative
and the absolute errors.
20. (i) Establish the general formula for absolute and relative errors for the function
v = f (u1 , u2 , . . . , un ) when absolute errors ui of each independent quantity ui
up uq ur
are known. Use this result for the function v = 1 s 2 t 3 to nd the upper bound
u4 u5
of the relative error.
(ii) Find the relative error in computing f (x) = 2x5 3x + 2 at x = 1, if the error
in x is 0.005.
1.42x + 3.45
(iii) If y =
here the coecients are roundedo, nd the absolute and
x + 0.75
relative errors in y when x = 0.5 0.1.
21. Given y = x4 y 5/2 , if x0 , y0 be the approximate values of x, y respectively and
x0 , y0 be the absolute errors in them, determine the relative error in u.
(a + b)c
, where a = 1.562 0.001, b = 10.3 0.02, c = 0.12
(d e)2
0.04, d = 10.541 0.004, e = 2.34 0.006. Find the absolute and the relative
errors in the result.
22. Calculate x =
23. (i) Determine the number of correct digits in the number 0.2318 if the relative
error is 0.3 101 .
(ii) Find the number of signicant gures in the approximate number 0.4785 given
that the relative error is 0.2 102 .
24. Find the smaller root of the equation x2 500x+1 = 0 using fourdigit arithmetic.
26 Numerical Analysis
25. Find the value of
103
Chapter 2
2.1
2.1.1
(2.1)
(2.2)
28 Numerical Analysis
Thus,
2 y0 = y1 y0 = (y2 y1 ) (y1 y0 ) = y2 2y1 + y0
2 y1 = y2 y1 = (y3 y2 ) (y2 y1 ) = y3 2y2 + y1
3 y0 = 2 y1 2 y0 = (y3 2y2 + y1 ) (y2 2y1 + y0 ) = y3 3y2 + 3y1 y0
3 y1 = y4 3y3 + 3y2 y1
and so on.
In general,
n+1 f (x) = [n f (x)], i.e., n+1 yi = [n yi ], n = 0, 1, 2, . . . .
(2.3)
(2.4)
y
y0
x1
y1
y0
2 y0
3 y0
y1
x2
2 y
y2
3 y
y2
x3
y3
x4
y4
4 y0
1
1
2 y2
y3
Backward dierences
(2.5)
29
i = n, n 1, . . . , 1.
(2.6)
That is,
y1 = y1 y0 , y2 = y2 y1 , . . . , yn = yn yn1 .
(2.7)
These dierences are called rst dierences. The second dierences are denoted by
2 y2 , 2 y3 , . . . , 2 yn . That is,
2 y2 = (y2 ) = (y2 y1 ) = y2 y1 = (y2 y1 ) (y1 y0 ) = y2 2y1 + y0 .
Similarly, 2 y3 = y3 2y2 + y1 , 2 y4 = y4 2y3 + y2 , and so on.
In general,
k yi = k1 yi k1 yi1 ,
i = n, n 1, . . . , k,
(2.8)
where 0 yi = yi , 1 yi = yi .
These backward dierences can be written in a tabular form and this table is known
as backward dierence or horizontal table.
Table 2.2 is the backward dierence table for the arguments x0 , x1 , . . . , x4 .
Table 2.2: Backward dierence table.
x
x0
x1
x2
x3
x4
2.1.3
y
y0
y1
y2
y3
y4
y1
y2
y3
y4
2 y2
2 y3
2 y4
3 y3
3 y4
4 y4
Central dierences
(2.9)
(2.10)
30 Numerical Analysis
where yi+1/2 = f (xi + h/2) and yi1/2 = f (xi h/2).
Thus y1/2 = y1 y0 , y3/2 = y2 y1 , . . . , yn1/2 = yn yn1 .
The second central dierences are
2 yi = yi+1/2 yi1/2 = (yi+1 yi ) (yi yi1 ) = yi+1 2yi + yi1 .
In general,
n yi = n1 yi+1/2 n1 yi1/2 .
(2.11)
y
y0
x1
y1
y1/2
2 y1
3 y3/2
y3/2
x2
2y
y2
3 y5/2
y5/2
x3
2y
y3
4 y2
y7/2
x4
y4
It is observed that all odd dierences have fraction suces and all the even dierences
are with integral suces.
2.1.4
Shift operator, E:
The shift operator is dened by
Ef (x) = f (x + h).
(2.12)
Eyi = yi+1 .
(2.13)
This gives,
That is, shift operator shifts the function value yi to the next higher value yi+1 .
The second shift operator gives
E 2 f (x) = E[Ef (x)] = E[f (x + h)] = f (x + 2h).
(2.14)
31
In general,
E n f (x) = f (x + nh) or E n yi = yi+nh .
(2.15)
(2.16)
and
E n f (x) = f (x nh).
(2.17)
(2.18)
f (x) =
i.e.,
(2.19)
Dierential operator, D:
The dierential operator is usually denoted by D, where
d
f (x) = f (x)
dx
d2
D2 f (x) = 2 f (x) = f (x).
dx
Df (x) =
2.1.5
(2.20)
Factorial notation
The factorial notation has many uses in calculus of nite dierence. This is used to
nd dierent dierences and antidierences. The nth factorial of x, denoted by x(n) ,
is dened by
x(n) = x(x h)(x 2h) (x n 1h),
where, each factor is decreased from the earlier by h; and x(0) = 1.
Similarly, the nth negative factorial of x is dened by
1
.
x(n) =
x(x + h)(x + 2h) (x + n 1h)
It may be noted that x(n) .x(n) = 1.
(2.21)
(2.22)
32 Numerical Analysis
2.2
f (x)
g(x)f (x) f (x)g(x)
Property 2.2.7
=
, g(x) = 0.
g(x)
g(x + h)g(x)
Proof.
f (x + h) f (x)
f (x)
=
g(x)
g(x + h)
g(x)
f (x + h)g(x) g(x + h)f (x)
=
g(x + h)g(x)
g(x)[f (x + h) f (x)] f (x)[g(x + h) g(x)]
=
g(x + h)g(x)
g(x)f (x) f (x)g(x)
.
=
g(x + h)g(x)
33
1
f (x)
=
.
f (x)
f (x + h)f (x)
Property 2.2.9 [cx ] = cx+h cx = cx (ch 1), for some constant c.
Property 2.2.10 [x Cr ] = xCr1 , where r is xed and h = 1.
Proof. [x Cr ] = x+1Cr xCr = xCr1 as h = 1.
Property 2.2.11 x(n) = nhx(n1) .
Proof.
x(n) = (x + h)(x + h h)(x + h 2h) (x + h n 1h)
x(x h)(x 2h) (x n 1h)
= x(x h)(x 2h) (x n 2h)[x + h {x (n 1)h}]
= nhx(n1) .
This property is analogous to the dierential formula D(xn ) = nxn1 when h = 1.
Most of the above formulae are similar to the corresponding formulae in dierential
calculus.
Property 2.2.12 The above formula can also be used to nd antidierence (like integration in integral calculus), as
1 x(n1) =
2.2.1
1 (n)
x .
nh
(2.23)
34 Numerical Analysis
Property 2.2.19 E
f (x)
g(x)
Ef (x)
.
Eg(x)
2.3
It is clear from the forward, backward and central dierence tables that in a denite
numerical case, the same values occur in the same positions, practically there are no
dierences among the values of the tables, but, dierent symbols have been used for the
theoretical importance.
Thus
yi = yi+1 yi = yi+1 = yi+1/2
2 yi = yi+2 2yi+1 + yi = 2 yi+2 = 2 yi+1
etc.
In general,
Again,
n yi = n yi+n ,
i = 0, 1, 2, . . . .
(2.24)
This relation indicates that the eect of the operator on f (x) is the same as that
of the operator E 1 on f (x). Thus
E1
Also,
or
E + 1.
(2.25)
That is,
1 E 1 .
(2.26)
The higher order forward dierence can be expressed in terms of the given function
values in the following way:
3 yi = (E 1)3 yi = (E 3 3E 2 + 3E 1)yi = y3 3y2 + 3y1 y0 .
35
There is a relation among the central dierence, , and the shift operator E, as
f (x) = f (x + h/2) f (x h/2) = E 1/2 f (x) E 1/2 f (x) = (E 1/2 E 1/2 )f (x).
That is,
E 1/2 E 1/2 .
(2.27)
1
f (x + h/2) + f (x h/2)
2
1
1
= E 1/2 f (x) + E 1/2 f (x) = (E 1/2 + E 1/2 )f (x).
2
2
Thus,
1
(2.28)
E 1/2 + E 1/2 .
2
The average operator can also be expressed in terms of the central dierence operator.
Again,
f (x) =
2
1 1/2
E + E 1/2 f (x)
4
1
1 1/2
= (E E 1/2 )2 + 4 f (x) = 2 + 4 f (x).
4
4
2 f (x) =
Hence,
1
1 + 2.
(2.29)
4
Some more relations among the operators , , E and are deduced in the following.
h2
h3
f (x) + f (x) +
2!
3!
h2 2 h3 3
= 1 + hD + D + D + f (x)
2!
3!
= ehD f (x).
(2.30)
36 Numerical Analysis
Hence,
Also,
E ehD .
(2.31)
hD log E.
(2.32)
This relation is used to separate the eect of E into that of the powers of and this
method of separation is called the method of separation of symbols.
The operators and can be expressed in terms of D, as shown below
1
1
f (x) = [E 1/2 + E 1/2 ]f (x) = ehD/2 + ehD/2 f (x)
2
2
hD
= cosh
f (x)
2
and f (x) = [E 1/2 E 1/2 ]f (x) = ehD/2 ehD/2 f (x)
hD
= 2 sinh
f (x).
2
Thus,
cosh
hD
Again,
2 cosh
2
hD
2
and 2 sinh
sinh
hD
2
hD
2
= sinh(hD).
(2.33)
(2.34)
(2.35)
is also useful.
Since E 1 + and E 1 1 , [from (2.25) and (2.26)]
from (2.32), it is obtained that
hD log E log(1 + ) log(1 ) sinh1 ().
(2.36)
1
f (x + 3h/2) + f (x + h/2) ,
2
while
Ef (x) = E
Hence,
1
f (x + h/2) + f (x h/2) = f (x + 3h/2) + f (x + h/2) .
2
2
1
E E.
(2.37)
37
2
2 2
1/2
+ 1+ ,
, (ii) E
+ , (iii)
(i) 1 + 1 +
2
2
2
4
E 1
+
(iv) (1 + )(1 ) 1, (v)
+ , (vi)
,
2
2
2
(vii) 2 .
Solution. (i) f (x) = 12 (E 1/2 + E 1/2 )(E 1/2 E 1/2 )f (x) = 12 [E E 1 ]f (x).
Therefore,
1
2 2
1 2
(1 + )f (x) = 1 + (E E ) f (x)
4
1 2
1
= 1 + (E 2 + E 2 ) f (x) = (E + E 1 )2 f (x)
4
4
2 2
1 1/2
1/2 2
) f (x) = 1 +
f (x).
= 1 + (E E
2
2
2 2
.
1+ 1+
2
Hence
2 2
(2.38)
1 1/2
1 1/2
1/2
1/2
f (x) =
[E + E
(ii) +
] + [E E
] f (x) = E 1/2 f (x).
2
2
2
Thus
E 1/2 + .
2
(iii)
2
+
2
(2.39)
2
f (x)
1+
4
1 1/2
1
1/2 2
1/2
1/2
= (E E
) f (x) + (E E
) 1 + (E 1/2 E 1/2 )2 f (x)
2
4
1
1
= [E + E 1 2]f (x) + (E 1/2 E 1/2 )(E 1/2 + E 1/2 )f (x)
2
2
1
1
= [E + E 1 2]f (x) + (E E 1 )f (x)
2
2
= (E 1)f (x).
Hence,
2
+
2
1+
2
E 1 .
4
(2.40)
38 Numerical Analysis
(iv) (1 + )(1 )f (x) = (1 + )[f (x) f (x) + f (x h)]
= (1 + )f (x h) = f (x h) + f (x) f (x h)
= f (x).
Therefore,
(1 + )(1 ) 1.
(v)
(2.41)
E 1
1
+
f (x) = [f (x h) + f (x)]
2
2
2
1
= [f (x) f (x h) + f (x + h) f (x)]
2
1
1
= [f (x + h) f (x h)] = [E E 1 ]f (x)
2
2
1 1/2
= (E + E 1/2 )(E 1/2 E 1/2 )f (x)
2
= f (x).
Hence
E 1
+
.
2
2
(vi)
(2.42)
+
1
f (x) = [f (x) + f (x)]
2
2
1
= [f (x + h) f (x) + f (x) f (x h)]
2
1
1
= [f (x + h) f (x h)] = [E E 1 ]f (x)
2
2
= f (x)
(as in previous case).
Thus,
+
.
2
(2.43)
(E 1/2 E 1/2 )2 2 .
(2.44)
39
The relations among the various operators are shown in Table 2.4.
Table 2.4: Relationship between the operators.
E
+1
(1 )1
E1
(1 )1 1
1 E 1
1 (1 + )1
2
(1 + )1/2
hD
log E
log(1 + )
log(1 )
hD
2
+ 1+
ehD
4
2
2
+ 1+
ehD 1
2
4
2
2
1 ehD
+ 1+
2
4
2 sinh(hD/2)
2
cosh(hD/2)
1+
4
1+
2 sinh1 (/2)
hD
f (x) = lim
Thus one can write,
f (x) hf (x).
Again,
f (x + h) f (x)
h0
h
f (x + h) f (x)
h
h
lim
h0
h
f (x + h) f (x)
2 f (x)
= lim
=
lim
.
h0
h0
h2
h2
f (x) = lim
2.4
40 Numerical Analysis
x(0)
x(1)
x(2)
x(3)
x(4)
=1
=x
= x(x h)
= x(x h)(x 2h)
= x(x h)(x 2h)(x 3h)
(2.45)
and so on.
The above relations show that x(n) , n = 1, 2, . . . is a polynomial of degree n in x.
Also, x, x2 , x3 , . . . can be expressed in terms of factorial notations x(1) , x(2) , x(3) , . . ., as
shown below.
1 = x(0)
x = x(1)
x2 = x(2) + hx(1)
x3 = x(3) + 3hx(2) + h2 x(1)
x4 = x(4) + 6hx(3) + 7h2 x(2) + h3 x(1)
(2.46)
and so on.
These relations show that xn can be expressed as a polynomial of x(1) , x(2) , . . . , x(n) ,
of degree n. Once a polynomial is expressed in a factorial notation, its dierences can
be obtained by using the formula like dierential calculus.
Example 2.4.1 Express f (x) = 2x4 + x3 5x2 + 8 in factorial notation and nd
its rst and second dierences.
Solution. Here we assume that h = 1.
Then by (2.46), x = x(1) , x2 = x(2) + x(1) , x3 = x(3) + 3x(2) + x(1) ,
x4 = x(4) + 6x(3) + 7x(2) + x(1) .
Using these values, the function f (x) becomes
f (x) = 2 x(4) + 6x(3) + 7x(2) + x(1) + x(3) + 3x(2) + x(1) 5 x(2) + x(1) + 8
= 2x(4) + 13x(3) + 12x(2) 2x(1) + 8.
Now, the Property 2.2.11, i.e., x(n) = nx(n1) is used to nd the dierences.
Therefore,
f (x) = 2.4x(3) + 13.3x(2) + 12.2x(1) 2.1x(0) = 8x(3) + 39x(2) + 24x(1) 2
and 2 f (x) = 24x(2) + 78x(1) + 24.
In terms of x,
f (x) = 8x(x 1)(x 2) + 39x(x 1) + 24x 2
and 2 f (x) = 24x(x 1) + 78x + 24.
From the relations of (2.46) one can conclude the following result.
41
Lemma 2.4.1 Any polynomial f (x) in x of degree n can be expressed in factorial notation with same degree, n.
This means, in conversion to the factorial notation, the degree of a polynomial remains
unchanged.
The above process to convert a polynomial in a factorial form is a labourious technique when the degree of the polynomial is large. The other systematic process, like
Maclaurins formula in dierential calculus, is used to convert a polynomial, even a
function, in factorial notation.
Let f (x) be a polynomial in x of degree n. In factorial notation, let it be
f (x) = a0 + a1 x(1) + a2 x(2) + + an x(n) ,
(2.47)
42 Numerical Analysis
f (x)
1
f (x)
2 f (x)
3 f (x)
4 f (x)
7
14
21
2
27
104
327
42
56
48
77
90
146
223
Thus using formula (2.48)
1
1
2
1
3
2
5
3
8
8
3
11
10
21
0
11
11
2.5
43
Dierence of a Polynomial
2)(n 3) 2 n4
h x
+b1 xn2 + (n 2)hxn3 +
+ + hn2 xn2
2!
+ + bn2 h
1
= b0 (n 1)hxn2 +
(n 1)(n 2)b0 h2 + (n 2)b1 h xn3 + + bn2 h
2!
= c0 xn2 + c1 xn3 + + cn3 x + cn2
where c0 = b0 (n 1)h, c1 = 2!1 (n 1)(n 2)b0 h2 + (n 2)b1 h, etc.
This expression shows that 2 f (x) is a polynomial of degree n 2.
It may be noted that the coecient of the leading term is c0 = b0 h(n 1) = n(n
1)h2 a0 and it is a constant quantity.
44 Numerical Analysis
In this way, one can nd n1 f (x) is a polynomial of degree one and let it be p0 x+p1 ,
i.e., n1 f (x) = p0 x + p1 .
Then n f (x) = p0 (x+h)+p1 p0 xp1 = p0 h, which is a constant. And n+1 f (x) =
0.
It can be shown that n f (x) = n(n 1)(n 2) 2 1 hn a0 = n!hn a0 .
Thus nally,
k f (x), k < n is a polynomial of degree n k,
n f (x) is constant, and
k f (x), k > n is zero.
Alternative proof.
It is observed that, a polynomial in x of degree n can be expressed as a polynomial in
factorial notation with same degree n.
Thus, if f (x) = a0 xn + a1 xn1 + a2 xn2 + + an1 x + an be the given polynomial
then it can be written as f (x) = b0 x(n) + b1 x(n1) + b2 x(n2) + + bn1 x(1) + bn .
Therefore,
f (x) = b0 nhx(n1) + b1 h(n 1)x(n2) + b2 h(n 2)x(n3) + + bn1 h.
Clearly this is a polynomial of degree n 1.
Similarly,
2 f (x) = b0 n(n 1)h2 x(n2) + b1 (n 1)(n 2)h2 x(n3) + + bn2 h2 ,
3 f (x) = b0 n(n 1)(n 2)h3 x(n3) + b1 (n 1)(n 2)(n 3)h3 x(n4)
+ + bn3 h3 .
In this way,
n f (x) = b0 n(n 1)(n 2) 2 1 hn x(nn)
= b0 n!hn ,
a constant quantity.
45
In this way,
n x(n) = n(n 1)(n 2) 2 1 hn x(nn) = n!h2 .
Example 2.5.1 Given xi = x0 + ih, i = 0, 1, 2, . . . , n; h > 0
and ui (x) = (x x0 )(x x1 ) (x xi ),
prove that
k ui (x) = (i + 1)i(i 1) (i k + 2)hk (x x0 )(x x1 ) (x xik ).
Solution. Here ui (x) = (x x0 )(x x1 ) (x xi ) = (x x0 )(i+1) (say).
Therefore,
ui (x) = (x + h x0 )(x + h x1 ) (x + h xi ) (x x0 ) (x xi )
= (x + h x0 )(x x0 )(x x1 ) (x xi1 )
(x x0 )(x x1 ) (x xi )
= (x x0 )(x x1 ) (x xi1 )[(x + h x0 ) (x xi )]
= (x x0 )(x x1 ) (x xi1 )(h + xi x0 )
[since xi = x0 + ih]
= (x x0 )(x x1 ) (x xi1 )(i + 1)h
= (i + 1)h(x x0 )(i) .
Similarly,
2 ui (x) = (i + 1)h[(x + h x0 )(x + h x1 ) (x + h xi1 )
(x x0 )(x x1 ) (x xi1 )]
= (i + 1)h(x x0 )(x x1 ) (x xi2 )[(x + h x0 ) (x xi1 )]
= (i + 1)h(x x0 )(i1) ih
= (i + 1)ih2 (x x0 )(i1) .
In similar way,
3 ui (x) = (i + 1)i(i 1)h3 (x x0 )(i2) .
Hence,
k ui (x) = (i + 1)i(i 1) (i k 2)hk (x x0 )(ik1)
= (i + 1)i(i 1) (i k + 2)hk (x x0 )(x x1 ) (x xik ).
2.6
Summation of Series
The nite dierence method is also used to nd the sum of a nite series. Two important
results are presented here.
46 Numerical Analysis
Theorem 2.1 If f (x) be dened only for integral values of independent variable x, then
b+h
b
f (x) = f (a) + f (a + h) + f (a + 2h) + + f (b) = F (x)
i.e.,
x=a
b
f (x) = F (x + h) F (x),
b = a + nh for some n
(2.49)
x=a
f (x) =
b
F (x) =
x=a
b
[F (x + h) F (x)]
x=a
Example 2.6.1 Use nite dierence method to nd the sum of the series
n
3
.
f (x), where f (x) = x(x 1) +
x(x + 1)(x + 2)
x=1
Solution. Let
F (x) =
f (x) =
x(x 1) +
3
x(x + 1)(x + 2)
x(2)
x(3)
+3
3
2
3
1
1
.
= x(x 1)(x 2)
3
2 x(x + 1)
= 1 x(2) + 31 x(3) =
Therefore,
n
x=1
3
1
3 1
1
0+
f (x) = [F (n + 1) F (1)] = (n + 1)n(n 1)
3
2 (n + 1)(n + 2)
2 1.2
1
1
3
3
= n(n2 1)
+ .
3
2 (n + 1)(n + 2) 4
47
Summation by parts
Like the formula integration by parts of integral calculus there is a similar formula in
nite dierence calculus. If f (x) and g(x) are two functions dened only for integral
values of x between a and b, then
b
b
b+h
f (x)g(x) = f (x)g(x)
x=a
x=a
n
x4x .
x=1
x4x = x.
4 .1
3 1
3
3
3
3
x=1
x=1
x=1
4n+1
4n+1 4 4 4(4n 1)
=n
.
= (n + 1)
3
3
3
4
3
n
2.7
x(1) =
x(2)
x(3) =
1
x(2)
=
.
x + 3h
(x + h)(x + 2h)(x + 3h)
(2.50)
48 Numerical Analysis
In this way,
x(n1) =
1
(x + h)(x + 2h) (x + n 1h)
1
(x + nh)(x + nh h)(x + nh 2h) (x + nh n 1h)
1
.
=
(x + nh)(n)
1 n+1
C1 + n+1 C2 2 + + n+1 u0
=
x2
x3
u2 + u3 +
2!
3!
x2 2
x3
= u0 + xEu0 + E u0 + E 3 u0 +
2!
3!
2
(xE)3
(xE)
+
+ u0
= 1 + xE +
2!
3!
u0 + xu1 +
= exE u0 = ex(1+) u0 = ex ex u0
(x)2 (x)3
+
+ u0
= ex 1 + x +
2!
3!
2
x 2
x3 3
x
= e u0 + xu0 + u0 + u0 +
2!
3!
1
1
u0 2 u0 3 u0
2 3
+ u0 = u0
+
+
=
1 +
2
2
4
8
2
4
8
16
fi = E f0 =
i
i
j=0
j f0 .
i
i
j=0
j f0 .
n
(1)i n Ci f [x + (n i)h],
i=0
where h is steplength.
Solution. We know that, E 1 and n (E 1)n .
n
(1)i n Ci E ni f (x).
Therefore, n f (x) = (E 1)n f (x) =
i=0
49
50 Numerical Analysis
Example 2.7.5 Find the polynomial f (x) which satises the following data and
hence nd the value of f (1.5).
x
f (x)
:
:
1
3
2
5
3
10
4
30
f (x)
3
f (x)
2 f (x)
3 f (x)
2
3
5
3
12
10
15
20
30
:
:
1
2
2
3
3
8
5
21
Solution. Here four values of f (x) are given. So, we consider f (x) be a polynomial
of degree 3. Thus the fourth dierences of f (x) vanish, i.e.,
4 f (x) = 0 or, (E 1)4 f (x) = 0
or, (E 4 4E 3 + 6E 2 4E + 1)f (x) = 0
or, E 4 f (x) 4E 3 f (x) + 6E 2 f (x) 4Ef (x) + f (x) = 0
51
or, f (x + 4) 4f (x + 3) + 6f (x + 2) 4f (x + 1) + f (x) = 0.
Here, h = 1 as the values are in spacing of 1 unit.
For x = 1 the above equation becomes
f (5) 4f (4) + 6f (3) 4f (2) + f (1) = 0 or, 21 4f (4) + 6 8 4 3 2 = 0
or, f (4) = 13.75.
Example 2.7.7 Use nite dierence method to nd the values of a and b in the
following table.
x
f (x)
:
:
0
5
2
a
4
8
6
b
8
20
10
32
Solution. Here, four values of f (x) are known, so we can assume that f (x) is a
polynomial of degree 3. Then, 4 f (x) = 0.
or, (E 1)4 f (x) = 0
or, E 4 f (x) 4E 3 f (x) + 6E 2 f (x) 4Ef (x) + f (x) = 0
or, f (x + 8) 4f (x + 6) + 6f (x + 4) 4f (x + 2) + f (x) = 0
[Here h = 2, because the values of x are given in 2 unit interval]
In this problem, two unknowns a and b are to be determined and needs two equations.
Therefore, the following equations are obtained by substituting x = 2 and x = 0 to
the above equation.
f (10) 4f (8) + 6f (6) 4f (4) + f (2) = 0 and
f (8) 4f (6) + 6f (4) 4f (2) + f (0) = 0.
These equations are simplies to
32 4 20 + 6b 4 8 + a = 0 and 20 4b + 6 8 4a 5 = 0.
That is, 6b + a 80 = 0 and 4b 4a + 63 = 0. Solution of these equations is
a = 2.9, b = 12.85.
Example 2.7.8 Find the value of
2 2
x .
E
Solution.
(E 1)2 2
2 2
x =
x
E
E
2
E 2E + 1 2
x
=
E
52 Numerical Analysis
f (x)
.
Example 2.7.9 Show that log f (x) = log 1
f (x)
Solution.
log f (x) = log f (x) log f (x h) = log f (x) log E 1 f (x)
E 1 f (x)
(1 )f (x)
f (x)
= log
= log
= log 1
E f (x)
f (x)
f (x)
f (x)
f (x) f (x)
= log 1
.
= log
f (x)
f (x)
The following formulae for antidierence can easily be veried, for h = 1.
(i) 1 cf (x) = c1 f (x), c being a constant.
(ii) 1 x(n) =
(iii) 1 ax =
1
(n+1) , n
n+1 x
1
ax ,
a1
(iv) 1 sin ax =
(v) 1 cos ax =
2.8
a = 1.
1
cos(ax a/2).
2 sin a/2
1
sin(ax a/2).
2 sin a/2
Dierence Equations
(2.51)
where a, b, c are constants, g(x) is a known function and f (x) is the unknown function.
The solution of a dierence equation is the value of the unknown function.
The dierence equation can also be expressed as a relation among the independent
variable and the successive values, i.e., f (x), f (x + h), f (x + 2h), . . . , of dependent
variable. For example, the dierence equation
22 f (x) f (x) + 5f (x) = x2 + 3x,
(2.52)
53
(2.53)
The dierence between the largest and the smallest arguments appearing in the difference equation with unit interval is called the order of the dierence equation.
The order of the equation (2.53) is (x + 2) x = 2, while the order of the equation
ux+3 8ux+1 + 5ux1 = x3 + 2 is (x + 3) (x 1) = 4. The order of the dierence
equation f (x + 2) 3f (x) = 0 is 3 as it is equivalent to ux+3 ux+2 3ux = 0.
A dierence equation in which ux , ux+1 , . . . , ux+n occur to the rst degree only and
there are no product terms is called linear dierence equation. Its general form is
a0 ux+n + a1 ux+n1 + + an ux = g(x).
(2.54)
If the coecients a0 , a1 , . . . , an are constants, then the equation (2.54) is called linear
dierence equation with constant coecients. If g(x) = 0 then the equation is
called homogenous otherwise it is called nonhomogeneous dierence equation.
The linear homogeneous equation with constant coecients of order n is
a0 ux+n + a1 ux+n1 + + an ux = 0.
(2.55)
or,
f (E)ux = 0,
(2.56)
54 Numerical Analysis
Let us consider another example of a sequence whose xth term is given by
ux = a2x + b3x
(2.57)
(2.58)
(2.59)
and
55
Hence
xn = xn1 + xn2 ,
n 3,
x1 = x2 = 1.
(2.60)
This is the dierence equation of the above stated problem and the solution is x1 =
1, x2 = 1, x2 = 2, x3 = 3, x4 = 5, x5 = 8, . . ., i.e., the sequence is {1, 1, 2, 3, 5, 8 . . .}
and this sequence is known as Fibonacci sequence.
Example 2.8.2 (The Tower of Hanoi). The Tower of Hanoi problem is a famous
problem of the late nineteenth century. The problem is stated below.
Let there be three pegs, numbered 1, 2, 3 and they are on a board and n discs of
dierence sizes with holes in their centres. Initially, these n discs are placed on one
peg, say peg 1, in order of decreasing size, with the largest disc at the bottom. The
rules of the puzzle are that the discs can be moved from one peg to another only one
at a time and no discs can be placed on the top of a smaller disc. The problem is to
transfer all the discs from peg 1 to another peg 2, in order of size, with the largest
disc at the bottom, in minimum number of moves.
Let xn be the number of moves required to solve the problem with n discs. If n = 1,
i.e., if there is only one disc on peg 1, we simply transfer it to peg 2 by one move.
Hence x1 = 1. Now, if n > 1, starting with n discs on peg 1 we can transfer the top
n 1 discs, following the rules of this problem to peg 3 by xn1 moves. During these
moves the largest disc at the bottom on peg 1 remains xed. Next, we use one move
to transfer the largest disc from peg 1 to peg 2, which was empty. Finally, we again
transfer the n 1 discs on peg 3 to peg 2 by xn1 moves, placing them on top of the
largest disc on peg 2 which remains xed during these moves. Thus, when n > 1,
(n 1) discs are transferred twice and one additional move is needed to move the
largest disc at the bottom from peg 1 to peg 2. Thus the recurrence relation is
xn = 2xn1 + 1 for n 2 and x1 = 1.
(2.61)
2.9
Several methods are used to solve dierence equations. Among them the widely used
methods are iterative method, solution using operators, solution using generating function, etc.
2.9.1
Iterative method
In this method the successive terms are substituted until the terms reduce to initial
term. The method is illustrated by example.
56 Numerical Analysis
n(n 1)
2
1
= 0 + n(n 1) [since x1 = 0]
2
1
= n(n 1).
2
Example 2.9.2 Solve the dierence equation for the Tower of Hanoi problem: xn =
2xn1 + 1, n 2 with x1 = 1.
Solution.
xn = 2xn1 + 1 = 2(2xn2 + 1) + 1
= 22 xn2 + (2 + 1) = 22 (2xn3 + 1) + (2 + 1)
= 23 xn3 + (22 + 2 + 1) = 23 (2xn4 + 1) + (22 + 2 + 1)
= 24 xn4 + (23 + 22 + 2 + 1).
2.9.2
This method is used to solve homogeneous as well as nonhomogeneous dierence equations. First we consider the homogeneous linear dierence equations with constant
coecients.
57
(2.62)
(2.63)
(2.64)
This equation is called the auxiliary equation (A.E.) for the dierence equation
(2.62). Since (2.64) is a quadratic equation, three types of roots may occur.
Case I. Let m1 and m2 be two distinct real roots of (2.64). In this case, the general
solution is ux = c1 mx1 + c2 mx2 , where c1 and c2 are arbitrary constants.
Case II. Let m1 , m1 be two real and equal roots of (2.64). In this case (c1 mx1 + c2 mx1 ) =
(c1 + c2 )mx1 = cmx1 is the only one solution of (2.62). To get the other solution (as a
second order dierence equation should have two independent solutions, like dierential
equation), let us consider ux = mx1 vx be its solution.
Since m1 , m1 are two equal roots of (2.64), the equation (2.63) may be written as
(E 2 2m1 E + m21 )ux = 0.
Substituting ux = mx1 vx to this equation, we obtain
x+2
x+2
mx+2
1 ux+2 2m1 vx+1 + m1 vx = 0
x+2
2
or, m1 (vx+2 2vx+1 + vx ) = 0 or, mx+2
1 vx = 0.
That is, 2 vx = 0. Since second dierence is zero, the rst dierence is constant and
hence vx is linear. Let vx = c1 + c2 x, where c1 , c2 are arbitrary constants.
Hence, in this case the general solution is
ux = (c1 + c2 x)mx1 .
Case III. If the roots m1 , m2 are complex, then m1 , m2 should be conjugate complex
and let them be ( + i) and ( i), where , are reals. Then the general solution
is
ux = c1 ( + i)x + c2 ( i)x .
To simplify the above expression, substituting = r cos , = r sin , where r =
2 + 2 and tan = /.
58 Numerical Analysis
Therefore,
ux = c1 rx (cos + i sin )x + c2 rx (cos i sin )x
= rx {c1 (cos x + i sin x) + c2 (cos x i sin x)}
= rx {(c1 + c2 ) cos x + i(c1 c2 ) sin x}
= rx (A cos x + B sin x), where A = c1 + c2 and B = i(c1 c2 ).
Example 2.9.3 Solve ux+1 8ux = 0.
Solution. This equation is written as (E 8)ux = 0. Let ux = cm2 be a solution.
The A.E. is m 8 = 0 or, m = 8.
Then ux = c8x , where c is an arbitrary constant, is the general solution.
Example 2.9.4 Solve the dierence equation ux = ux1 +ux2 , x 2, u0 = 1, u1 =
1. Also, nd the approximate value of ux when x tends to a large number.
x
2
Solution. Let
ux = cm be a solution. The A.E. is m m 1 = 0
1 5
.
or, m =
2
Therefore, general solution is
1+ 5 x
1 5 x
u x = c1
+ c2
,
2
2
1+ 5
1 5
+ c2
.
1 = c1 + c2 and 1 = c1
2
2
Solution of these equations is
5+1
1 5
and c2 = .
c1 =
2 5
2 5
Hence, the particular solution is
1 + 5 x+1
1
1 5 x+1
ux =
.
2
2
5
1 5 x+1
0 and therefore,
When x then
2
1 1 + 5 x+1
.
ux
2
5
59
1 x
1 x
a =
a , provided f (a) = 0.
f (E)
f (a)
1 x
1
a (x) = ax
(x), where f (aE) = 0.
f (E)
f (aE)
60 Numerical Analysis
Case V. g(x) = sin ax or cos ax.
1 iax
1
sin ax = Imaginary part of
e .
(i) P.I. =
f (E)
f (E)
(ii) P.I. =
1 iax
1
cos ax = Real part of
e .
f (E)
f (E)
P.I. =
E2
i
7
.
Then A.E. is m2 m + 2 = 0 or, m =
2
1
7
Here the roots are complex. Let = r cos and
= r sin .
2
2
Therefore, r = 2, tan = 7.
1
(x2 + 5x )
E+2
1
1
{x(2) + x(1) } +
5x
2
(1 + ) (1 + ) + 2
25 5 + 2
1
5x
(2)
(1)
{x
+
x
}
+
2 + + 2
22
1
2
1
+
5x
1+
{x(2) + x(1) } +
2
2
22
2
2
2
+
+
1
5x
1
+
{x(2) + x(1) } +
2
2
2
22
x
5
1 (2)
1
1
x + x(1) (2 + 2x(1) + 1) + (2) +
2
2
4
22
5x
5x
1 2
1 (2)
[x 1] +
= (x x 1) + .
2
22
2
22
E2
1
5x
ux2 = ( 2)x [c1 cos x + c2 sin x] + (x2 x 1) + ,
2
22
where
= 7 and c1 , c2 are arbitrary constants.
x
Example 2.9.8 Show that the solution of the equation ux+2 + ux = 2 sin
2
x
x
+ x sin
.
is given by ux = a cos
2
2
Solution. Let ux = cmx be a solution of ux+2 + ux = 0.
Then A.E. is m2 + 1 = 0 or, m = i.
Therefore, C.F. is A(i)x + B(i)x .
Substituting 0 = r cos , 1 = r sin , where r = 1, = /2.
61
62 Numerical Analysis
Then C.F. reduces to
A{r(cos + i sin )}x + B{r(cos i sin )}x
x
x
x
x
+ i sin
i sin
= A cos
+ B cos
2
2
2
2
x
x
(B A)i sin
= (A + B) cos
2
2
x
x
= a cos cos
a sin sin
, where A + B = a cos , (B A)i = a sin
2
2
x
+ .
= a cos
2
x
1
1
2 sin
= Imaginary part of 2
2eix/2
+1
2
E +1
1
1
[since (ei/2 )2 + 1 = 0]
= I.P. of 2eix/2 i/2 2
(e
E) + 1
1
1
= I.P. of 2eix/2 i 2
1
1 = I.P. of 2eix/2
e E +1
1 (1 + )2
ix/2 1
1 + 1
= I.P. of 2e
2
2
1
1
1 = I.P. of 2eix/2
x
= I.P. of 2eix/2
2
2
x
x
= I.P. of (x) cos
+ i sin
2
2
x
= x sin
.
2
x
x
+ x sin
.
Therefore, the general solution is ux = a cos
2
2
P.I. =
E2
E2
1
1
3n =
3n = 3n .
4E + 4
(E 2)2
63
or,
(E 5)f (n) = 3.
2.9.3
11 n1 3
.
5
4
4
Generating function
an xn
(2.65)
n=0
64 Numerical Analysis
un xn = u0 +
n=0
That is,
G(x) 2 =
un xn = 2 +
n=1
un xn .
n=1
un xn .
n=1
un xn = 2
n=1
That is,
G(x) 2 = 2x
= 2x
n=1
un1 xn + 3
n=1
un1 xn1 + 3
xn .
n=1
xn
n=1
un xn + 3
xn 1
n=0
= 2x G(x) + 3
n=0
1
1
1x
since
xn =
n=0
1
1x
3
1.
1x
Therefore, the generating function for this dierence equation or for the sequence
{un } is
Thus,
(1 2x)G(x) =
3
1
(1 x)(1 2x) 1 2x
3
5
n
n
=5
(2x) 3
x =
(5.2n 3)xn .
G(x) =
n=0
n=0
n=0
65
n
an x = 2 + 3x +
an xn .
G(x) = a0 + a1 x +
That is,
n=2
n=2
an x = G(x) 2 3x.
n
n=2
n=2
an xn 5
an1 xn + 6
n=2
or, G(x) 2 3x 5x
an2 xn = 0
n=2
n=2
or, G(x) 2 3x 5x
an xn a0
an2 xn2 = 0
n=2
+ 6x2 G(x) = 0
n=0
2 7x
. This is the generating function for the given dier1 5x + 6x2
B
(A + B) (3A + 2B)x
2 7x
A
+
=
.
=
2
1 5x + 6x
1 2x 1 3x
(1 2x)(1 3x)
.
whose solution is A = 3, B = 1. Thus, G(x) =
1 2x 1 3x
Now,
G(x) = 3(1 2x)1 (1 3x)1
=3
(2x)n
(3x)n =
(3.2n 3n )xn .
n=0
n=0
n=0
66 Numerical Analysis
2.10
Exercise
1. Dene the operators: forward dierence (), backward dierence (), shift (E),
central dierence () and average ().
2. Prove the following relations among the operators
2 2
2 2
, (ii) E 1/2 +
(i) 1 + 1 +
2
2
+
E 1
(iii)
, (iv)
+ ,
2
2
2
2
(v)
+ 1 + ( 2 /4), (vi) hD sinh1 (),
2
1
hD
(vii) hD
log(1 +
) log(1 ) sinh (), (viii) E e ,
i=0
2n
i=0
67
2 x Eex
e 2 x
(ii) e =
E
e
f (x)
(iii) log f (x) = log 1 +
f (x)
2
(iv)
x3 = 6x.
E
7. Prove that, if the spacing h is very small then the forward dierence operator is
almost equal to dierential operator, i.e., for small h, n f (x) hn Dn f (x).
8. Show that the operators , , E, and are commute with one another.
9. Express 3 yi and 4 y4 in terms of y.
10. Prove the following relations:
(i) ux = ux1 + ux2 + 2 ux3 + + n1 uxn + n uxn1
(ii) u1 + u2 + u3 + + un =n C1 u0 +n C2 u0 +n C3 2 u0 + + n1 u0
(iii) n yx = yn+x n C1 yx+n1 +n C2 yx+n2 + (1)n yx
(iv) u1 x + u2 x2 + u3 x3 +
x
x2
x3
=
u1 +
u
+
2 u1 + .
1
1x
(1 x)2
(1 x)3
11. Show that
(i) [f (x)g(x)] = f (x)g(x) + g(x + h)f (x)
(ii) n f (x) = n f (x + nh), where n is a positive integer
(iii) f (x) = f (x) f (x)
(iv) f (x) + f (x) = (/)f (x) (/)f (x).
12. The nth dierence n be dened as n f (x) = n1 f (x+h)n1 f (x), (n 1).
If f (x) is a polynomial of degree n show that f (x) is a polynomial of degree n1.
Hence deduce that the nth dierence of f (x) is a constant.
13. If r (x) = (x x0 )(x x1 ) (x xr ) where xr = x0 + rh, r = 0, 1, 2, . . . , n,
calculate k r (x).
14. If fi is the value of f (x) at xi where xi = x0 + ih, i = 1, 2, . . . prove that
fi = E i f0 =
i
i
j=0
j f0 .
68 Numerical Analysis
16. Taking h = 1, compute the second, third and fourth dierences of f (x) = 3x4
2x2 + 5x 1.
17. Construct the forward dierence table for the following tabulated values of f (x)
and hence nd the values of 2 f (3), 3 f (2), 4 f (0).
x
f (x)
:
:
0
4
1
7
2
10
3
20
4
45
5
57
6
70
18. Use nite dierence method to nd a polynomial which takes the following values:
x
f (x)
:
:
2
12
1
6
0
0
1
6
2
10
:
:
0
12
2
6
4
0
6
?
8
25
20. Use nite dierence method to nd the value of f (2.2) from the following data.
x
f (x)
:
:
1
3
2
24
3
99
4
288
5
675
69
1
is x [A(1 + 5)x + B(1 5)x ] (x2 + 6x + 13).
2
2
28. Show that
the solution of ux+2
+ a 2ux = cos ax is
x
x
a cos ax + cos a(2 x)
.
+ B sin
+
ux = ax A cos
2
2
1 + 2a2 cos 2a + a4
where a = 3 17.
30. The seeds of a certain plant when one year old produce eighteen fold. A seed is
planted and every seed subsequently produced is planted as soon as it is produced.
Prove that the number of grain at the end of nth year is
1
un =
a
11 + a
2
n+1
11 a
2
n+1
where a = 3 17.
31. The rst term of a sequence {un } is 1, the second is 4 and every other term is the
arithmetic mean of the two preceding terms. Find un and show that un tends to
a denite limit as n .
32. The rst term of a sequence is 1, the second is 2 and every term is the sum of the
two proceeding terms. Find the nth term.
70 Numerical Analysis
33. If ur satises the dierence equation ur 4ur1 + ur2 = 0, 2 r n, where
A sinh(n r)
.
un = 0 and u0 = A, show that, if = log(2 + 3) then ur =
sinh n
34. Show that the general solution of the national income equation
1
1
yn+2 yn+1 yn = nh + A where h, A are constants, is given by
2n
4
yn = c1 m1 + c2 mn2 + 4hn + 4(A 6h) where c1 , c2 are arbitrary constants and the
values of m1 and m2 you are to actually nd out. Also show that yn /n tends to
nite limit as n .
35. Use generating functions to solve the following dierence equations.
(i) xn = 3xn1 , n 1 and x0 = 2.
(ii) xn = 5xn1 + 2n , n 1 and x0 = 2.
(iii) xn = xn1 + n for n 1 and x0 = 1.
(iv) un un1 un2 = 0 for n 2 and u0 = 0, u1 = 1.
(v) an + an2 = 2an1 , n 2 and a0 = 0, a1 = 1.
Chapter 3
Interpolation
Sometimes we have to compute the value of the dependent variable for a given independent variable, but the explicit relation between them is not known. For example,
the Indian population are known to us for the years 1951, 1961, 1971, 1981, 1991 and
2001. There is no exact mathematical expression available which will give the population for any given year. So one can not determine the population of India in the year
2000 analytically. But, using interpolation one can determine the population (obviously
approximate) for any year.
The general interpolation problem is stated below:
Let y = f (x) be a function whose analytic expression is not known, but a table of
values of y is known only at a set of values x0 , x1 , x2 , . . ., xn of x. There is no other
information available about the function f (x). That is,
f (xi ) = yi , i = 0, 1, . . . , n.
(3.1)
The problem of interpolation is to nd the value of y(= f (x)) for an argument, say,
x . The value of y at x is not available in the table.
A large number of dierent techniques are used to determine the value of y at x = x .
But one common step is nd an approximate function, say, (x), corresponding to
the given function f (x) depending on the tabulated value. The approximate function
should be simple and easy to handle. The function (x) may be a polynomial, exponential, geometric function, Taylors series, Fourier series, etc. When the function (x) is
a polynomial, then the corresponding interpolation is called polynomial interpolation.
The polynomial interpolation is widely used interpolation technique, because, polynomials are continuous and can be dierentiated and integrated term by term within any
range.
A polynomial (x) is called interpolating polynomial if yi = f (xi ) = (xi ), i =
dk f
dk
0, 1, 2, . . . , n and
=
for some nite k, and x is one of the values of x0 ,
dxk x
dxk x
71
72 Numerical Analysis
y
y = f (x) +
y = (x)
:
y=f (x)
y = f (x)
x
x0 x1
x2
xn
3.1
Let y = f (x) be a real valued function dened on an interval [a, b]. Let x0 , x1 , . . . , xn be
n + 1 distinct points in the interval [a, b] and y0 , y1 , . . . , yn be the corresponding values
of y at these points, i.e., yi = f (xi ), i = 0, 1, . . . , n, are given.
Interpolation
73
(3.2)
The polynomial (x) is called the interpolation polynomial and the points xi ,
i = 0, 1, . . . , n are called interpolation points.
Let the polynomial (x) be of the form
(x) =
n
Li (x) yi ,
(3.3)
i=0
where each Li (x) is polynomial in x, of degree less than or equal to n, called the
Lagrangian function.
The polynomial (x) satises the equation (3.2) if
0, for i = j
Li (xj ) =
1, for i = j.
That is, the polynomial Li (x) vanishes only at the points x0 , x1 , . . . , xi1 , xi+1 , . . . , xn .
So it should be of the form
Li (x) = ai (x x0 )(x x1 ) (x xi1 )(x xi+1 ) (x xn ),
where ai , a constant whose value is determined by using the relation
Li (xi ) = 1.
Then ai (xi x0 )(xi x1 ) (xi xi1 )(xi xi+1 ) (xi xn ) = 1.
or, ai = 1/{(xi x0 )(xi x1 ) (xi xi1 )(xi xi+1 ) (xi xn )}.
Therefore,
Li (x) =
n
Li (x) yi ,
i=0
n
x xj
.
xi xj
j=0
j=i
(3.4)
74 Numerical Analysis
In this notation, the polynomial (x) is
n
n
x xj
yi .
(x) =
xi xj
j=0
i=0
j=i
(3.5)
w(x)
.
(x xi )w (xi )
n
i=0
w(x)
yi .
(x xi )w (xi )
(3.6)
Now,
(x + 1)(x 1)
x2 1
(x x1 )(x x2 )
=
=
.
(x0 x1 )(x0 x2 )
(2 + 1)(2 1)
3
(x + 2)(x 1)
x2 + x 2
(x x0 )(x x2 )
L1 (x) =
=
=
.
(x1 x0 )(x1 x2 )
(1 + 2)(1 1)
2
(x + 2)(x + 1)
x2 + 3x + 2
(x x0 )(x x1 )
=
=
.
L2 (x) =
(x2 x0 )(x2 x1 )
(1 + 2)(1 + 1)
6
L0 (x) =
Interpolation
75
Therefore,
x2 + x 2
x2 + 3x + 2
x2 1
(5) +
(1) +
1
3
2
6
= 1 + x x2 .
f (x)
Thus, f (0) = 1.
The Lagrangian coecients can be computed from the following scheme. The dierences are computed, rowwise, as shown below:
x x0
x1 x0
x2 x0
xn x0
x0 x1
x x1
x2 x1
xn x1
x0 x2
x1 x2
x x2
x x2
x0 xn
x1 xn
x2 xn
x xn
From this table, it is observed that the product of diagonal elements is w(x). The
product of the elements of rst row is (x x0 )w (x0 ), product of elements of second row
is (x x1 )w (x1 ) and so on. Then the Lagrangian coecient can be computed using
the formula
w(x)
.
Li (x) =
(x xi )w (xi )
Linear Lagrangian Interpolation
Let x0 and x1 be two points and y0 and y1 be the corresponding values of y. In this
case,
(x) = L0 (x)y0 + L1 (x)y1
x x1
x x0
x x0
=
y0 +
y1 = y0 +
(y1 y0 ).
x0 x1
x1 x0
x1 x0
(3.7)
76 Numerical Analysis
Also,
w (xi ) = (xi x0 )(xi x1 ) (xi xi1 )(xi xi+1 )(xi xi+2 ) (xi xn )
= (ih)(i 1)h (i i 1)h(i i + 1)h(i i + 2)h (i n)h
= hn i(i 1) 1 (1)(2) ({(n i)}
= hn i!(1)ni (n i)!.
Using these values, the relation (3.5) becomes
(x) =
=
n
hn+1 s(s 1)(s 2) (s n)
i=0
n
i=0
yi
s(s 1)(s 2) (s n)
yi ,
i!(n i)!(s i)
(3.8)
where x = x0 + sh.
For given tabulated values, the Lagranges interpolation polynomial exists and unique.
These are proved in the following theorem.
Theorem 3.2 The Lagranges interpolation polynomial exists and unique.
Proof. The Lagranges interpolation formula satised the condition
yi = (xi ), i = 0, 1, . . . , n.
(3.9)
For n = 1,
(x) =
x x0
x x1
y0 +
y1 .
x0 x1
x1 x0
(3.10)
For n = 2,
(x) =
(x x0 )(x x2 )
(x x1 )(x x2 )
y0 +
y1
(x0 x1 )(x0 x2 )
(x1 x0 )(x1 x2 )
(x x0 )(x x1 )
y2 .
+
(x2 x0 )(x2 x1 )
(3.11)
n
i=0
Li (x)yi ,
(3.12)
Interpolation
77
where
Li (x) =
(3.13)
Expression (3.10) is a linear function, i.e., a polynomial of degree one and also,
(x0 ) = y0 and (x1 ) = y1 .
Also, expression (3.11) is a second degree polynomial and (x0 ) = y0 , (x1 ) =
y1 , (x2 ) = y2 , i.e., satisfy (3.9). Thus, the condition (3.13) for n = 1, 2 is fullled.
The functions (3.13) expressed in the form of a fraction whose numerator is a polynomial of degree n and whose denominator is a nonzero number. Also, Li (xi ) = 1 and
Li (xj ) = 0 for j = i, j = 0, 1, . . . , n. That is, (xi ) = yi . Thus, the conditions of (3.9)
are satised. Hence, the Lagranges polynomial exists.
Uniqueness of the polynomial
Let (x) be a polynomial of degree n, where
(xi ) = yi , i = 0, 1, . . . , n.
(3.14)
(3.15)
(3.16)
If (x) (x) = 0, then this dierence is a polynomial of degree at most n and it has
at most n zeros, which contradicts (3.16), whose number of zeros is n + 1. Consequently,
(x) = (x). Thus (x) is unique.
3.2
78 Numerical Analysis
Therefore,
w(x) = (x x0 )(x x1 ) (x xn )
= an+1 (z z0 )(z z1 ) (z zn ).
w (xi ) = (xi x0 )(xi x1 ) (xi xi1 )(xi xi+1 ) (xi xn )
= an (zi z0 )(zi z1 ) (zi zi1 )(zi zi+1 ) (zi zn ).
Thus,
Li (x) =
w(x)
(x xi )w (xi )
an+1 (z z0 )(z z1 ) (z zn )
a(z zi )an (zi z0 )(zi z1 ) (zi zi1 )(zi zi+1 ) (zi zn )
w(z)
= Li (z).
=
(z zi )w (zi )
=
n
Li (x) = 1.
i=0
Li (x) =
n
i=0
w(x)
(x xi )w (xi )
(3.17)
(3.19)
Interpolation
A1 =
1
w (x1 )
79
1
1
and An =
.
w (xi )
w (xn )
Using these results, equation (3.18) becomes
Also, Ai =
1
1
1
=
+
+
w(x)
(x x0 )w (x0 ) (x x1 )w (x1 )
1
1
+ +
+
(x xi )w (xi )
(x xn )w (xn )
n
w(x)
.
i.e., 1 =
(x xi )w (xi )
i=0
3.3
Li (x) =
n
i=0
w(x)
= 1.
(x xi )w (xi )
(3.20)
f (n+1) ()
,
(n + 1)!
(3.21)
where I.
Proof. Let the error En (x) = f (x) (x), where (x) is a polynomial of degree less
than or equal to n, which approximates the function f (x).
Now, En (xi ) = f (xi ) (xi ) = 0 for i = 0, 1, . . . , n.
By virtue of the above result, it is assumed that En (x) = w(x)k, where w(x) =
(x x0 )(x x1 ) . . . (x xn ).
The error at any point, say, x = t, other than x0 , x1 , . . . , xn is En (t) = w(t)k
or, f (t) (t) = kw(t).
(3.22)
80 Numerical Analysis
Let us construct an auxiliary function
F (x) = f (x) (x) kw(x).
(3.23)
f (n+1) ()
.
(n + 1)!
f (n+1) ()
.
(n + 1)!
f (n+1) ()
.
(n + 1)!
Note 3.3.1 The above expression gives the error at any point x.
But practically it has little utility, because, in many cases f (n+1) () cannot be determined.
If Mn+1 be the upper bound of f (n+1) () in I, i.e., if f (n+1) () Mn+1 in I then
the upper bound of En (x) is
En (x)
Mn+1
w(x).
(n + 1)!
(3.24)
f (n+1) ()
.
(n + 1)!
(3.25)
Interpolation
81
Note 3.3.3 (Error bounds for equally spaced points, particular cases)
Assume that, f (x) is dened on [a, b] that contains the equally spaced points. Suppose,
f (x) and the derivatives up to n + 1 order are continuous and bounded on the intervals
[x0 , x1 ], [x0 , x2 ] and [x0 , x3 ] respectively. That is, f (n+1)()  Mn+1 for x0 xn ,
for n = 1, 2, 3. Then
h2 M2
,
8
h3 M3
(ii) E2 (x) ,
9 3
h4 M4
(iii) E3 (x)
,
24
(i) E1 (x)
x0 x x1
(3.26)
x0 x x2
(3.27)
x0 x x3 .
(3.28)
f (2) ()
.
2!
Let g1 (s) = s(s 1). g1 (s) = 2s 1. Then s = 1/2, which is the solution of g1 (s) = 0.
The extreme value of g1 (s) is 1/4.
Therefore,
h2 M2
1 M2
=
.
E1 (x) h2
4
2!
8
f (3) ()
.
3!
1
Let g2 (s) = s(s 1)(s 2). Then g2 (s) = 3s2 6s + 2. At g2 (s) = 0, s = 1 .
3
Again g2 (s) = 6(s 1) < 0 at s = 1 13 .
Therefore, the maximum value of g2 (s) is
(ii) E2 (x) = s(s 1)(s 2)h3
2
1
1
1
1 = .
1
3
3 3
3
3
Thus,
h3 M3
2
M3
E2 (x) h3
= .
6
3 3
9 3
f (4 ()
.
4!
Let g3 (s) = s(s 1)(s 2)(s 3). Then g3 (s) = 4s3 18s2 + 22s 6.
i.e., 2s3 9s2 + 11s 3 = 0.
At extrema, g3 (s) = 0,
3 3 5
.
This gives s = ,
2
2
(iii) E3 (x) = s(s 1)(s 2)(s 3)h4
82 Numerical Analysis
g3 (s)
g3 (3/2)
g3
3 5
> 0.
18s + 11. Then
< 0 and
2
3
3 5
9
But, g3 (s) = 1 at s =
and g3 (s) =
at x = .
2
16
2
Therefore the maximum value of g3 (s) is 1.
Hence,
h4 M4
M4
=
.
E3 (x) 1.h4
24
24
=
6s2
= 0.28125.
E1 (x)
8
8
For quadratic polynomial the spacing of the points is h = (1.5 0)/2 = 0.75 and its
error bound is
E2 (x)
(0.75)3 0.997495
h3 M5
= 0.0269955.
9 3
9 3
Interpolation
83
The spacing for cubic polynomial is h = (1.5 0)/3 = 0.5 and thus the error bound
is
(0.5)4 1.0
h4 M4
= 0.0026042.
E3 (x)
24
24
Example 3.3.2 A function f (x) dened on the interval (0, 1) is such that f (0) =
0, f (1/2) = 1, f (1) = 0. Find the quadratic polynomial p(x) which agrees with f (x)
for x = 0, 1/2, 1.
d3 f
1
for 0 x 1.
If 3 1 for 0 x 1, show that f (x) p(x)
dx
12
Solution. Given x0 = 0, x1 = 1/2, x2 = 1 and f (0) = 0, f (1/2) = 1, f (1) = 0.
From Lagranges interpolating formula, the required quadratic polynomial is
(x x0 )(x x2 )
(x x1 )(x x2 )
f (x0 ) +
f (x1 )
(x0 x1 )(x0 x2 )
(x1 x0 )(x1 x2 )
(x x0 )(x x1 )
f (x2 )
+
(x2 x0 )(x2 x1 )
(x 0)(x 1)
(x 0)(x 1/2)
(x 1/2)(x 1)
0+
(1) +
0
=
(0 1/2)(0 1)
(1/2 0)(1/2 1)
(1 0)(1 1/2)
= 4x(x 1).
p(x) =
f ()
E(x) = (x x0 )(x x1 )(x x2 )
3!
f ()
or, E(x) = x x0 x x1 x x2 
3!
d3 f
1
as 3 1 in 0 x 1 .
x 0x 1/2x 11.
3!
dx
Now, x 0 1, x 1/2 1/2 and x 1 1 in 0 x 1.
1
1 1
Hence, E(x) 1. .1. = .
2 6
12
1
That is, f (x) p(x) .
12
Example 3.3.3 Determine the step size h (and number of points n) to be used in
the tabulation of f (x) = cos x in the interval [1, 2] so that the quadratic interpolation
will be correct to six decimal places.
84 Numerical Analysis
Solution. The upper bound of error in quadratic polynomial is
E2 (x)
f (x) = cos x,
h3 M3
,
9 3
f (x) = sin x,
f (x) = cos x,
1x2
h3
Hence 1 5 106 ,
9 3
This gives h 0.0427 and n =
1x2
i.e., h3 45 3 106 .
21
= 23.42 24.
h
(x) =
Interpolation
85
y
y=f (x)
y=(x)
 x
Figure 3.2: The graph of the function f (x) = ex and the polynomial
(x) = 0.309636243x2 0.941756802x + 1.
86 Numerical Analysis
#include <stdio.h>
#include <math.h>
void main()
{
int n, i, j; float xg, x[20], y[20], sum=0, prod=1;
printf("Enter the value of n and the data
in the form x[i],y[i] ");
scanf("%d",&n);
for(i=0;i<=n;i++) scanf("%f %f",&x[i],&y[i]);
printf("\nEnter the interpolating point x ");
scanf("%f",&xg);
for(i=0;i<=n;i++)
{
prod=1;
for(j=0;j<=n;j++)
{
if(i!=j) prod*=(xgx[j])/(x[i]x[j]);
}
sum+=y[i]*prod;
}
printf("\nThe given data is ");
for(i=0;i<=n;i++) printf("\n(%6.4f,%6.4f)",x[i],y[i]);
printf("\nThe value of y at x= %5.2f is %8.5f ", xg, sum);
} /* main */
A sample of input/output:
Enter the value of n and the data in the form x[i],y[i] 4
1
5
1.5 8.2
2
9.2
3.2 11
4.5 16
Enter the interpolating point x 1.75
The given data is
(1.0000,5.0000)
(1.5000,8.2000)
(2.0000,9.2000)
(3.2000,11.0000)
(4.5000,16.0000)
The value of y at x= 1.75 is 8.85925
Interpolation
3.4
87
Finite Dierences
Dierent types of nite dierences are introduced in Chapter 2. Some of them are
recapitulated here.
Let a function y = f (x) be known as (xi , yi ) at (n+1) points xi , i = 0, 1, . . . , n, where
xi s are equally spaced, i.e., xi = x0 + ih, h is the spacing between any two successive
points xi s. That is, yi = f (xi ), i = 0, 1, . . . , n.
3.4.1
Forward dierences
(3.29)
(3.30)
It is observed that dierence of any order can easily be expressed in terms of the
ordinates yi s with binomial coecients.
All orders forward dierences can be written in a tabular form shown in Table 3.1.
This dierence table is called forward dierence table or diagonal dierence
table.
88 Numerical Analysis
y
y0
2 y
3 y
4 y
y0
x1
2 y0
y1
3 y0
y1
x2
2 y1
y2
y2
x3
y3
x4
y4
4 y0
3 y
2 y2
y3
3.4.2
Backward dierences
(3.31)
Table 3.2 shows how the backward dierences of all orders can be formed.
The backward dierence table is sometimes called horizontal dierence table.
3.4.3
If there is an error in any entry among the tabulated values of a function, then this
error propagates to other entries of higher order dierences. To illustrate the behaviour
of propagation of error, we assume that an error is present in the number, say, y3 .
Interpolation
89
y
y0
y1
y2
y3
y4
2 y
3 y
4 y
y1
y2
y3
y4
2 y2
2 y3
2 y4
3 y3
3 y4
4 y4
y
y0
x1
y1
2 y
3 y
4 y
5 y
y0
2 y0
3 y0 +
y1
x2
2 y
y2
y2 +
x3
x4
y3 +
y3
3 y
2 y2 2
x6
y6
4 y0 4
5 y0 + 10
4 y1 + 6
3 y
3 y
y4
y5
1
2
+ 3
2 y3 +
y4
x5
4 y2 4
5 y1 10
2 y4
y5
Table 3.3 shows the propagation of error in a dierence table and how the error aects
the dierences. From this table, the following observations are noted.
(i) The eect of the error increases with the order of the dierences.
(ii) The error is maximum (in magnitude) along the horizontal line through the erroneous tabulated value.
(iii) The second dierence column has the errors , 2, , in the third dierence column, the errors are , 3, 3, . In the fourth dierence column the expected
errors , 4, 6, 4, (this column is not sucient to show all of the expected errors). Thus, in the pth dierence column, the coecients of errors are the binomial
coecients in the expansion of (1 x)p .
90 Numerical Analysis
(iv) The algebraic sum of errors in any column (complete) is zero. If there is any error
in a single entry of a table, then from the dierence table one can detect and
correct such error.
Detection of errors using dierence table
Dierence table may be used to detect errors in a set of tabular values. From Table
3.3, it follows that if an error is present in a given data, the dierences of some order
will become alternating in sign. Thus, higher order dierences should be formed till the
error is revealed.
To detect the position of the error in an entry, the following steps may be proceed.
(i) Form the dierence table. If at any stage, the dierences do not follow a smooth
pattern, then one can conclude that there is an error.
(ii) If the dierences of some order (it is generally happens in higher order) becomes
alternating in sign then the middle entry has an error.
Let y = f (x) be a function whose explicit form is unknown. But, the values of y at
the equispaced points x0 , x1 , . . . , xn , i.e., yi = f (xi ), i = 0, 1, . . . , n are known. Since
x0 , x1 , . . . , xn are equispaced then xi = x0 + ih, i = 0, 1, . . . , n, where h is the spacing.
It is required to construct a polynomial (x) of degree less than or equal to n satisfying
the conditions
yi = (xi ),
i = 0, 1, . . . , n.
(3.32)
Since (x) is a polynomial of degree at most n, so (x) can be taken in the following
form
(x) = a0 + a1 (x x0 ) + a2 (x x0 )(x x1 ) + a3 (x x0 )(x x1 )(x x2 )
+ + an (x x0 )(x x1 ) (x xn1 ),
(3.33)
Interpolation
91
(x) = y0 + (x x0 )
(3.34)
(x) = y0 + (uh)
x x0
.
where u =
h
This is known as Newton or NewtonGregory forward dierence interpolating
polynomial.
Example 3.5.1 The following table gives the values of ex for certain equidistant
values of x. Find the value of ex when x = 0.612 using Newtons forward dierence
formulae.
x
y
:
:
0.61
1.840431
0.62
1.858928
0.63
1.877610
0.64
1.896481
0.65
1.915541
92 Numerical Analysis
Solution. The forward dierence table is
x
0.61
y
1.840431
0.62
1.858928
0.63
1.877610
0.64
1.896481
2 y
3 y
0.018497
0.000185
0.018682
0.000004
0.000189
0.018871
0.0
0.000189
0.019060
0.65
1.915541
0.612 0.61
x x0
=
= 0.2.
h
0.01
u(u 1) 2
u(u 1)(u 2) 3
y0 +
y0
2!
3!
0.2(0.2 1)
0.000185
= 1.840431 + 0.2 0.018497 +
2
0.2(0.2 1)(0.2 2)
0.000004
+
6
= 1.840431 + 0.003699 0.000015 + 0.00000019
= 1.844115.
y(0.612) = y0 + uy0 +
3.5.1
f (n+1) ()
(n + 1)!
f (n+1) ()
(using x = x0 + uh)
(n + 1)!
y0 .
E(x)
(n + 1)!
Interpolation
93
A particular case:
If 0 < u < 1 then
2
1
1
1
u(u 1) = (1 u)u = u u =
u and
4
2
4
(u 2)(u 3) (u n) (2)(3) (n) = n!.
Then,
1
1 n!
n+1 y0  =
n+1 y0 .
E(x)
4 (n + 1)!
4(n + 1)
2
94 Numerical Analysis
Program 3.2
.
/* Program Newton Forward Interpolation
This program finds the value of y=f(x) at a given x when
the function is supplied as (x[i],y[i]), i=0, 1, ..., n.
Assumed that xs are equispaced.*/
#include<stdio.h>
#include<math.h>
void main()
{
int i,j,n; float x[20],y[20],xg,sum,prod=1,u,dy[20],h;
printf("Enter number of subintervals ");
scanf("%d",&n);
printf("Enter x and y values ");
for(i=0;i<=n;i++) scanf("%f %f",&x[i],&y[i]);
printf("Enter interpolating point x ");
scanf("%f",&xg);
h=x[1]x[0];
u=(xgx[0])/h;
for(j=0;j<=n;j++) dy[j]=y[j];
prod=1; sum=y[0];
for(i=1;i<=n;i++)
{
for(j=0;j<=ni;j++) dy[j]=dy[j+1]dy[j];
prod*=(ui+1)/i;
sum+=prod*dy[0];
}
printf("The value of y at x=%f is %f ",xg,sum);
}
A sample of input/output:
Enter number of subintervals 4
Enter x and y values
140 3.685
150 5.854
160 6.302
170 8.072
180 10.225
Enter interpolating point x 142
The value of y at x=142.000000 is 4.537069
Interpolation
3.6
95
(3.36)
(3.37)
yn yn1
yn
=
.
h
h
3 yn
4 yn
n yn
,
a
=
,
.
.
.
,
a
=
.
4
n
3!h3
4!h4
n!hn
When the values of ai s are substituted in (3.36) then the polynomial (x) becomes
yn
2 yn
+ (x xn )(x xn1 )
h
2!h2
3
yn
+(x xn )(x xn1 )(x xn2 )
+
3!h3
n yn
.
+(x xn )(x xn1 )(x xn2 ) (x x1 )
n!hn
(x) = yn + (x xn )
(3.38)
96 Numerical Analysis
Using above results, (3.38) becomes
2 yn
yn
3 yn
+ vh(v + 1)h
+
vh(v
+
1)h(v
+
2)h
+
h
2!h2
3!h3
n yn
+vh(v + 1)h(v + 2)h (v + n 1)h
n!hn
v(v + 1) 2
v(v + 1)(v + 2) 3
yn +
yn +
= yn + vyn +
2!
3!
v(v + 1)(v + 2) (v + n 1) n
yn .
(3.39)
+
n!
(x) = yn + vh
:
:
0.20
1.6596
0.22
1.6698
0.24
1.6804
0.26
1.6912
0.28
1.7024
0.30
1.7139
f (x)
1.6596
1.6698
1.6804
1.6912
1.7024
1.7139
f (x)
2 f (x)
3 f (x)
0.0102
0.0106
0.0108
0.0112
0.0115
0.0004
0.0002
0.0004
0.0003
0.0002
0.0002
0.0001
0.29 0.30
x xn
=
= 0.5.
h
0.02
v(v + 1) 2
v(v + 1)(v + 2) 3
f (xn )+
f (xn )+
2!
3!
0.5(0.5 + 1)
0.0003
= 1.7139 0.5 0.0115 +
2
0.5(0.5 + 1)(0.5 + 2)
(0.0001)
+
6
= 1.7139 0.00575 0.0000375 + 0.00000625
= 1.70811875 1.7081.
Interpolation
97
Example 3.6.2 The population of a town in decennial census were given in the
following table.
Year
Population (in thousand)
:
:
1921
46
1931
66
1941
81
1951
93
1961
101
Estimate the population for the year 1955 using Newtons backward and forward
formulae and compare the results.
Solution.
Using Newtons backward formula
The backward dierence table is
Population
(y)
46
66
81
93
101
2 y
3 y
4 y
20
15
12
8
5
3
4
2
1
xxn
h
Year
(x)
1921
1931
1941
1951
1961
19551961
10
= 0.6.
v(v + 1) 2
v(v + 1)(v + 2) 3
yn +
yn
2!
3!
v(v + 1)(v + 2)(v + 3) 4
yn
+
4!
0.6(0.6 + 1)
(4)
= 101 0.6 8 +
2
0.6(0.6 + 1)(0.6 + 2)
(1)
+
6
0.6(0.6 + 1)(0.6 + 2)(0.6 + 3)
(3)
+
24
= 96.8368 97.
y(1955) = yn + vyn +
:
:
1961
101
1951
93
1941
81
1931
66
1921
46
98 Numerical Analysis
The forward dierence table is
Year
x
1961
Population
y
101
1951
93
y
8
12
1941
81
15
1931
66
20
1921
2 y
3 y
4 y
4
1
3
5
46
1955 1961
x x0
=
= 0.6.
h
10
u(u 1) 2
u(u 1)(u 2) 3
y0 +
y0
2!
3!
u(u 1)(u 2)(u 3) 4
y0
+
4!
0.6(0.6 1)(0.6 2)
0.6(0.6 1)
(4) +
1
= 101 + 0.6 (8) +
2
6
0.6(0.6 1)(0.6 2)(0.6 3)
(3)
+
24
= 101 4.8 + 0.48 + 0.056 + 0.1008
= 96.8368 97.
y(1955) = y0 + uy0 +
Therefore the population of the town in the year 1955 was 97 thousand and this result
is same with the result obtained by Newtons backward dierence formula.
3.6.1
f (n+1) ()
(n + 1)!
f (n+1) ()
,
(n + 1)!
(3.40)
x xn
and lies between min{x0 , x1 , . . . , xn , x} and max{x0 , x1 , . . . , xn , x}.
h
Interpolation
99
Note 3.6.1 The Newtons backward dierence interpolation formula is used to compute
the value of f (x) when x is near to xn , i.e., when x is at the end of the table.
Newtons forward and Newtons backward formulae does not give accurate value of f (x)
when x is in the middle of the table. To get more accurate result another formula may
be used. There are several methods available to solve this type of problem. Among
them Gaussian forward and backward, Stirlings and Bessels interpolation formulae are
widely used.
3.7.1
(3.41)
i.e., a2 =
(x) = y0 + (x x0 )
1)h
+
h
2!h2
3!h3
2n1 y(n1)
+(s + n 1)h sh(s 1)h (s n 1)h
(2n 1)!h2n1
2n yn
+(s + n 1)h sh(s 1)h (s n 1)h(s n)h
(2n)!h2n
(x) = y0 + sh
(3.43)
Interpolation
3 y1
2 y1
+ (s + 1)s(s 1)
+
2!
3!
2n1 y(n1)
+(s + n 1) s(s 1) (s n 1)
(2n 1)!
2n yn
+(s + n 1) s(s 1) (s n 1)(s n)
(2n)!
2
3
y1
y1
+ s(s2 12 )
= y0 + sy0 + s(s 1)
2!
3!
4y
2
+
+s(s2 12 )(s 2)
4!
2n1 y(n1)
2
2
+s(s2 n 1 )(s2 n 2 ) (s2 12 )
(2n 1)!
2n yn
2
2
.
+s(s2 n 1 )(s2 n 2 ) (s2 12 )(s n)
(2n)!
101
= y0 + sy0 + s(s 1)
(3.44)
(3.45)
x x0
and lies between min{xn , x(n1) , . . . , x0 , x1 , . . . , xn1 , xn }
h
and max{xn , x(n1) , . . . , x0 , x1 , . . . , xn1 , xn }.
In case of 2n arguments, the error is
where s =
f 2n ()
,
(2n)!
f 2n ()
(2n)!
(3.47)
(3.48)
Interpolation
103
(3.49)
The coecients ai s are unknown constants. These values are determined by using
the relations (3.48). Substituting x = x0 , x1 , x1 , x2 , x2 , . . . , xn , xn to (3.49) in succession. Note that xi xj = (i + j)h and (xi xj ) = (i + j)h. Then it is found
that
y0 = a0
(x1 ) = a0 + a1 (x1 x0 )
i.e., y1 = y0 + a1 (h),
y0 y1
y1
i.e., a1 =
=
h
h
(x1 ) = a0 + a1 (x1 x0 ) + a2 (x1 x1 )(x1 x0 )
y1
+ a2 (2h)(h)
y1 = y0 + h.
h
y1 y0 (y0 y1 )
2 y1
i.e., a2 =
=
2!h2
2!h2
(x2 ) = a0 + a1 (x2 x0 ) + a2 (x2 x1 )(x2 x0 )
+a3 (x2 x1 )(x2 x0 )(x2 x1 )
2 y1
y1
(2h) +
i.e., y2 = y0 +
(h)(2h) + a3 (h)(2h)(3h)
h
2!h2
= y0 2(y0 y1 ) + (y1 2y0 + y1 ) + a3 (1)3 (3!)h3
y1 3y0 + 3y1 y2
3 y2
or, a3 =
=
.
3!h3
3!h3
In this manner, the other values are obtained as
a4 =
4 y2
5 y3
2n1 yn
2n yn
,
a
=
,
.
.
.
,
a
=
,
a
=
.
5
2n1
2n
4!h4
5!h5
(2n 1)!h2n1
(2n)!h2n
y1
2 y1
+ (x x1 )(x x0 )
1!h
2!h2
3
y2
+(x x1 )(x x0 )(x x1 )
3!h3
4 y2
+(x x2 )(x x1 )(x x0 )(x x1 )
+
4!h4
(x) = y0 + (x x0 )
2n1 yn
(2n 1)!h2n1
2n yn
+(x xn )(x x1 )(x x0 )(x x1 ) (x xn1 )
.
(3.50)
(2n)!h2n
As in previous case, a new unit less variable s is introduced to reduce the above
x x0
i.e, x = x0 + sh.
formula into a simple form, where s =
h
Then
x x0 ih
x xi
=
= s i and
h
h
x x0 + ih
x xi
=
= s + i, i = 0, 1, 2, . . . , n.
h
h
Then the above formula is transferred to
(s + 1)s 2
(s + 1)s(s 1) 3
y1 +
y2
2!
3!
(s + 2)(s + 1)s(s 1) 4
y2 +
+
4!
(s + n 1) (s + 1)s(s 1) (s n + 1) 2n1
yn
+
(2n 1)!
(s + n)(s + n 1) (s + 1)s(s 1) (s n + 1) 2n
yn .
+
(2n)!
(x) = y0 + sy1 +
(3.51)
Interpolation
105
yn ,
+
(2n 1)!
(x) = y0 + sy1 +
where s =
3.7.4
(3.52)
x x0
.
h
3.8
(x) =
(3.54)
3.9
This central dierence formula is also obtained by taking the arithmetic mean of Gausss
forward and backward interpolation formulae. But, one dierence is that the backward
formula taken after one modication.
Let us consider 2n equispaced points x(n1) , . . . , x1 , x0 , x1 , . . . , xn1 , xn as arguments, where xi = x0 ih, h is the spacing.
If x0 , y0 be the initial values of x and y respectively, then the Gausss backward
dierence interpolation formula (3.52) is
s(s + 1) 2
(s + 1)s(s 1) 3
y1 +
y2
2!
3!
(s + 2)(s + 1)s(s 1) 4
y2 +
+
4!
(s + n 1) (s + 1)s(s 1) (s n + 1) 2n1
yn .
+
(2n 1)!
(x) = y0 + sy1 +
(3.55)
Interpolation
107
yn+1 . (3.56)
+
(2n 1)!
1 (x) = y1 + (s 1)y0 +
Taking arithmetic mean of (3.56) and Gausss forward interpolation formula given by
(3.45),
1 (x) + (x)forward
2
y 0 + y1
1
s(s 1) 2 y0 + 2 y1
=
+ s
y0 +
2
2
2!
2
1
(s 2 )s(s 1) 3
s(s 1)(s + 1)(s 2) 4 y2 + 4 y1
+
y1 +
3!
4!
2
1
(s 2 )s(s 1)(s + 1)(s 2) 5
y2 +
+
5!
(s 12 )s(s 1)(s + 1) (s + n 2)(s n 1) 2n1
+
y(n1) , (3.57)
(2n 1)!
(x) =
x x0
.
h
x x0 1
1
and then the above formula reduces to
Introducing u = s =
2
h
2
where s =
(x) =
u2 14 2 y1 + 2 y0 u(u2 14 ) 3
y0 + y 1
+ uy0 +
+
y1
2
2!
2
3!
2
(u2 14 )(u2 94 ) (u2 (2n3)
) 2n1
4
+
y(n1) .
(2n 1)!
(3.58)
The formulae given by (3.57) and (3.58) are known as Bessels central dierence
interpolation formula.
Example 3.9.1 Use the central dierence interpolation formula of Stirling or Bessel
to nd the values of y at (i) x = 1.40 and (ii) x = 1.60 from the following table
x
y
:
:
1.0
1.0000
1.25
1.0772
1.50
1.1447
1.75
1.2051
2.00
2.2599
xi
1.00
yi
1.0000
1.25
1.0772
1.50
1.1447
1.75
1.2051
yi
2 yi
3 yi
0.0772
0.0097
0.0675
0.0026
0.0071
0.0604
0.0015
0.0056
0.0548
2
2.00
1.2599
y(1.40) =
Interpolation
109
y(1.60) = y0 + s
3.10
5 y2
..
.
= 4 y1 4 y2 ,
..
..
.
.
(x) = y0 + s(y1 y0 ) +
+
y(n2)
(2n 1)!
s(s 1) (s + 1)s(s 1)
+ (1 s)y0 +
2 y1
2!
3!
(s + 1)s(s 1)(s 2) (s + 2)(s + 1)s(s 1)(s 2)
4 y2 +
+
4!
5!
y(n1)
+
(2n 1)!
s(s2 12 ) 2
s(s2 11 )(s2 22 ) 4
y0 +
y1 +
= sy1 +
3!
5!
+
y(n2)
(2n 1)!
u(u2 11 ) 2
u(u2 12 )(u2 22 ) 4
y1 +
y2 +
+ uy0 +
3!
5!
y(n1) ,
(3.59)
(2n 1)!
where s =
x x0
and u = 1 s.
h
:
:
1.0
1.0000
1.25
1.1180
1.50
1.2247
1.75
1.3229
2.00
1.4142
2.25
1.5000
xi
1.00
yi
1.0000
1.25
1.1180
yi
2 yi
3 yi
4 yi
0.1180
0.0113
0.1067
0
1.50
0.0028
0.0085
1.2247
0.0982
1.75
0.0069
1.3229
0.0913
2.00
2.25
1.5000
0.0002
0.0014
0.0055
1.4142
0.0858
0.0012
0.0016
Interpolation
111
s(s2 12 ) 2
s(s2 12 )(s2 22 ) 4
y(1.60) = sy1 +
y0 +
y1
3!
5!
u(u2 12 ) 2
u(u2 12 )(u2 22 ) 4
y1 +
y2
+ uy0 +
3!
5!
0.4(0.16 1)
= 0.4 1.3229 +
(0.0069)
6
0.4(0.16 1)(0.16 4)
(0.0002) + 0.6 1.2247
+
120
0.6(0.36 1)(0.36 4)
0.6(0.36 1)
(0.0085) +
(0.0012)
+
6
120
= 0.5292 + 0.0004 0.0000025 + 0.7348 + 0.0005 0.00001
= 1.2649.
3.10.1
These two formulae are closely related and one formula can be deduced from the other
one. Now, starting from Bessels formula (3.57):
y 0 + y1
1
s(s 1) 2 y0 + 2 y1
+ s
y0 +
2
2!
2
2
1
s 2 s(s 1)
x x0
3 y1 + , where s =
.
+
3!
h
y 0 + y1
1
s(s 1) 2 y0 + 2 y1
=
+ s
(y1 y0 ) +
2
2!
2
2
1
s 2 s(s 1)
(2 y0 2 y1 ) +
+
3!
s(s 1) s(s 1)(s 12 ) 2
y1 +
= (1 s)y0 +
4
6
s(s 1) s(s 1)(s 12 ) 2
+sy1 +
+
y0 +
4
6
s(s2 12 ) 2
u(u2 12 ) 2
= sy1 +
y0 + + uy0 +
y1 +
3!
3!
where u = 1 s.
(x) =
x3
y3
2 y
3 y
4 y
y3
x2
y2
2 y
3
33
y2
x1
y1
Newtons Backward
2 y2
4 y3
3 y2
y1
x0
x1
y1
?
R
 3  6  Bessels Formula
6 y0  6
? y1
?
?
R
R 3
y0
y1
x2
y2
y3
2 y1
4 y0
3 y1
y2
x3
4 y1
2 y0
Newtons Forward
2 y2
Interpolation
113
last interpolating point is taken as initial point x0 . But, in central dierence interpolation formulae, a middle point is taken as the initial point x0 .
3.11
Newtons interpolation formula generates successively higher order interpolation formula. The Aitken interpolation formula served the same purpose. But it has one
advantage that it can be easily programmed for a computer.
Let y = f (x) be given for n + 1 distinct points x0 , x1 , . . . , xn , i.e., yi = f (xi ), i =
0, 1, . . . , n are given, where the points xi , i = 0, 1, . . . , n need not be equispaced. To
compute the value of y for a given x the iterations proceed as follows:
to nd the value of y obtain a rst approximation by taking rst two points; then obtain
its second approximation by taking the rst approximations and so on.
The linear polynomial for the points x0 and x1 is
x x1
x x0
1
y0 +
y1 =
[(x1 x)y0 (x0 x)y1 ]
x0 x1
x1 x0
x1 x0
y0 x0 x
1
.
=
x1 x0 y1 x1 x
p01 (x) =
(3.60)
In general,
1
p0j (x) =
xj x0
y0 x0 x
yj xj x , j = 1, 2, . . . , n.
(3.61)
Here p0j (x) is a polynomial of degree less than or equal to 1, for the points x0 and
xj .
The polynomial
p01 (x) x1 x
1
, j = 2, 3, . . . , n.
p01j (x) =
(3.62)
xj x1 p0j (x) xj x
is a polynomial of degree less than or equal to 2.
The polynomial p01j (x) interpolates the points x0 , x1 and xj .
In general, the polynomial for the (k + 1) points x0 , x1 , . . . , xk and xj is
p012k (x) xk x
1
, j = k + 1, . . . , n.
p012kj (x) =
xj xk p012(k1)j (x) xj x
(3.63)
xn
yj
y0
y1
y2
y3
y4
yn
p0j
p01j
p012j
p0123j
p01
p02
p03
p04
p0n
p012
p013
p014
p01n
p0123
p0124
p012n
p01234
p0123n
xj x
x0 x
x1 x
x2 x
x3 x
x4 x
xn x
Example 3.11.1 Find the value of y(1.52) by iterated linear interpolation using
the following table.
x
y(x)
:
:
1.4
1.8330
1.5
1.9365
1.6
2.0396
p0j
p01j
p012j
1.9572
1.9570
1.9568
1.9572
1.9572
1.9572
1.7
2.1424
yj
1.8330
1.9365
2.0396
2.1424
Now
p0j =
p01 =
p02 =
p03 =
p01j =
p012 =
p013 =
xj x
0.12
0.02
0.08
0.18
y0 x0 x
1
,
j = 1, 2, 3.
xj x0 yj xj x
1 1.8330 0.12
= 1.9572.
0.1 1.9365 0.02
1 1.8330 0.12
= 1.9570.
0.2 2.0396 0.08
1 1.8330 0.12
= 1.9568.
0.3 2.1424 0.18
p01 x1 x
1
,
j = 2, 3.
xj x1 p0j xj x
1 1.9572 0.02
= 1.9572.
0.1 1.9570 0.08
1 1.9572 0.02
= 1.9572.
0.2 1.9568 0.18
Interpolation
p012j
p0123
115
p012 x2 x
1
,
=
j = 3.
xj x2 p01j xj x
1 1.9572 0.08
=
= 1.9572.
0.1 1.9572 0.18
3.12
Interpolation
117
f [x0 , x1 ] f [x1 , x2 ]
.
x0 x2
f [xi , xj ] f [xj , xk ]
In general, f [xi , xj , xk ] =
.
xi xk
f [x0 , x1 , x2 ] =
f (x0 ) f (x1 )
cc
=
= 0.
x0 x1
x0 x1
g(x0 ) g(x1 )
cf (x0 ) cf (x1 )
f (x0 ) f (x1 )
=
=c
= cf [x0 , x1 ].
x0 x1
x0 x1
x0 x1
h[x0 , x1 ] =
f (x0 ) f (x1 )
f (x1 ) f (x0 )
=
= f [x1 , x0 ].
x0 x1
x1 x0
1
1
f (x0 ) +
f (x1 ).
x0 x1
x1 x0
f (x1 ) +
f (x2 )
x1 x2
x2 x1
1
f (x0 )
=
(x0 x2 )(x0 x1 )
1
1
f (x1 ) +
f (x2 ).
+
(x1 x0 )(x1 x2 )
(x2 x0 )(x2 x1 )
f [x0 , x1 , x2 ] =
Interpolation
119
f [x0 , x1 , . . . , xn ] =
i=0
(3.64)
From these relations it is easy to observe that the divided dierences are symmetric.
5. For equispaced arguments, the divided dierences can be expressed in terms of
forward dierences.
That is,
f [x0 , x1 , . . . , xn ] =
1
hn .n!
n y0 .
=
x0 x2
2h h
h
2
y0
y1 y0
=
.
=
2
2h
2!h2
f [x0 , x1 ] =
(xk+1 x0 ) k! hk
k! hk
1
[k y1 k y0 ]
=
(k + 1)k! hk+1
k+1 y0
=
.
(k + 1)! hk+1
f [x0 , x1 , . . . , xk , xk+1 ] =
1
hn .n!
n y0 .
(3.65)
6. Divided dierences for equal arguments or divided dierences for conuent arguments.
If the arguments are equal then the divided dierences have a meaning. If two
arguments are equal then the divided dierence has no meaning as denominator
becomes zero. But, by limiting process one can dene the divided dierences for
equal arguments which is known as conuent divided dierences.
f [x0 , x0 ] = lim f [x0 , x0 + ] = lim
0
f (x0 + ) f (x0 )
= f (x0 ),
= lim
f (x0 )
0
f (x0 +)f (x0 )
f [x0 , x0 ] f [x0 , x0 + ]
f (x0 ) f (x0 + ) + f (x0 ) 0
form
= lim
0
2
0
f (x0 ) f (x0 + )
= lim
(by LHospital rule)
0
2
f (x0 )
=
.
2!
0
f (x0 )
.
3!
In general,
(k+1) times
f k (x0 )
f [x0 , x0 , . . . , x0 ] =
.
k!
In other words,
(3.66)
(k+1) times
dk
f (x0 ) = k! f [x0 , x0 , . . . , x0 ].
k
dx
(3.67)
Interpolation
121
Then
f (x) f (x0 )
x x0
xn xn0
xn1 x0n1
xn2 x0n2
x x0
= a0
+ a1
+ a2
+ + an1
x x0
x x0
x x0
x x0
= a0 [xn1 + xn2 x0 + xn3 x20 + + xx0n2 + x0n1 ]
f [x, x0 ] =
f [x, x0 , x1 ] =
3.13
f (x) f (x0 )
.
x x0
(3.68)
This formula is known as Newtons fundamental or Newtons general interpolation formula including error term.
Tabular form of divided dierences are shown in Table 3.4.
Table 3.4: Divided dierence table.
x
x0
x0 x1
x0 x2
x1
x1 x2
x1 x3
x2
x2 x3
x3
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
f (x)
First
Second
Third
f (x0 )
f [x0 , x1 ]
f (x1 )
f [x0 , x1 , x2 ]
f [x1 , x2 ]
f (x2 )
f [x1 , x2 , x3 ]
f [x2 , x3 ]
f (x3 )
f [x0 , x1 , x2 , x3 ]
Interpolation
123
Error term
The error term
E(x) = (x x0 )(x x1 ) (x xn )f [x0 , x1 , . . . , xn , x]
f n+1 ()
, [using (3.66)]
= (x x0 )(x x1 ) (x xn )
(n + 1)!
where min{x0 , x1 , . . . , xn , x} < < max{x0 , x1 , . . . , xn , x}.
Example 3.13.1 Find the value of y when x = 1.5 from the following table:
x
y
:
:
1
0.6931
5
1.7918
7
2.0794
10
2.3979
12
2.5649
2nd
3rd
y
0.6931
1.7918
4
6
9
11
0.0218
0.1438
2.0794
3
5
4th
0.2747
1st
0.0016
0.0075
0.1062
10
2.3979
0.0001
0.0004
0.0045
0.0835
12
2.5649
(3.69)
3.14
3.14.1
f (x0 )
2 f (x0 )
+ (x x0 )(x x1 )
1!h
2!h2
3
f (x0 )
+ (x x0 )(x x1 )(x x2 )
+
3!h3
n f (x0 )
+ E(x),
+ (x x0 )(x x1 ) (x xn1 )
n!hn
(x) = f (x0 ) + (x x0 )
where
E(x) = (x x0 )(x x1 ) (x xn )f [x, x0 , x1 , . . . , xn ]
n+1 f ()
= (x x0 )(x x1 ) (x xn )
.
(n + 1)! hn+1
Now, a unit less quantity u =
formula.
So, x xi = (u i)h.
Then
x x0
, i.e., x = x0 + uh is introduced to simplify the
h
u(u 1) 2
f (x0 ) +
2!
u(u 1)(u 2) (u n + 1) n
f (x0 ) + E(x),
+
n!
f n+1 ()
,
E(x) = u(u 1)(u 2) (u n)
(n + 1)!
min{x, x0 , x1 , . . . , xn } < < max{x, x0 , x1 , . . . , xn }.
(x) = f (x0 ) + uf (x0 ) +
3.14.2
Let
(x) = f (xn ) + (x xn )f [xn , xn1 ] + (x xn )(x xn1 )f [xn , xn1 , xn2 ]
+ + (x xn )(x xn1 ) (x x1 )f [xn , xn1 , . . . , x1 , x0 ]
+E(x),
Interpolation
where
E(x) = (x xn )(x xn1 ) (x x1 )(x x0 )f [x, xn , xn1 , . . . , x1 , x0 ].
From the relation (3.65), we have
f [xn , xn1 , . . . , xnk ] =
k f (xnk )
k f (xn )
=
.
k!hk
k!hk
Therefore,
f (xn )
2 f (xn )
+ (x xn )(x xn1 )
+
1!h
2!h2
n f (xn )
+ E(x),
+ (x xn )(x xn1 ) (x x1 )(x x0 )
n!hn
(x) = f (xn ) + (x xn )
where
E(x) = (x xn )(x xn1 ) (x x1 )(x x0 )
n+1 f ()
,
(n + 1)!hn+1
n
i=0
f (xi )
.
(xi x0 )(xi x1 ) (xi xi1 )(xi xi+1 ) (xi xn )
For (n + 2) arguments x, x0 , . . . , xn ,
f [x, x0 , x1 , . . . , xn ] =
+
f (x)
(x x0 )(x x1 ) (x xn )
n
f (xi )
i=0
f (x)
f (xi )
+
,
w(x)
(xi x)w (xi )
n
i=0
n
i=0
n
i=0
w(x)f (xi )
+ w(x)f [x, x0 , x1 , . . . , xn ]
(x xi )w (xi )
Li (x)f (xi ) + w(x)
f n+1 ()
[using (3.65)]
(n + 1)!
125
3.15
In the following, it is proved that the Lagranges interpolation formula and Newtons
divided dierence interpolation formula are equivalent.
The Lagranges interpolation polynomial for the points (xi , yi ), i = 0, 1, . . . , n of
degree n is
(x) =
n
Li (x)yi ,
(3.70)
i=0
where
Li (x) =
(3.71)
f (x0 )
f (x1 )
+
= f (x0 ) + (x x0 )
x0 x1 x1 x0
+(x x0 )(x x1 )
f (x1 )
f (x2 )
f (x0 )
+
+
(x0 x1 )(x0 x2 ) (x1 x0 )(x1 x2 ) (x2 x0 )(x2 x1 )
+ + (x x0 )(x x1 ) (x xn1 )
f (xn )
f (x0 )
+ +
.
(3.72)
(x0 x1 ) (x0 xn )
(xn x0 ) (xn xn1 )
Interpolation
127
(x x0 )(x x2 ) (x xn1 )
+ +
(x0 x2 )(x0 x3 ) (x0 xn )
(x x0 )(x x3 ) (x xn1 )
(x x1 )(x x2 )
x x0
+ +
1+
(x0 x1 )(x0 x2 )
x0 x2
(x0 x3 )(x0 x4 ) (x0 xn )
x x0
(x x1 )(x x2 ) (x xn1 )
1+
(x0 x1 )(x0 x2 ) (x0 xn1 )
x0 xn
(x x1 )(x x2 ) (x xn1 )(x xn )
(x0 x1 )(x0 x2 ) (x0 xn1 )(x0 xn )
L0 (x).
1+
=
=
=
=
Similarly, it can be shown that the coecient of f (x1 ) is L1 (x), coecient of f (x2 )
is L2 (x) and so on.
Thus, (3.72) becomes
(x) = L0 (x)f (x0 ) + L1 (x)f (x1 ) + + Ln (x)f (xn ) =
n
Li (x)f (xi ).
i=1
Thus the Lagranges interpolation and Newtons divided dierence interpolation formulae are equivalent.
Example 3.15.1 A function y = f (x) is given at the points x = x0 , x1 , x2 . Show
that the Newtons divided dierence interpolation formula and the corresponding
Lagranges interpolation formula are identical.
Solution. The Newtons divided dierence formulae is given as
y = f (x) = y0 + (x x0 )f [x0 , x1 ] + (x x0 )(x x1 )f [x0 , x1 , x2 ]
f (x1 ) f (x0 )
+ (x x0 )(x x1 )
= y0 + (x x0 )
x1 x0
f (x0 )
f (x1 )
f (x2 )
+
+
(x x0 )(x x1 )
(x0 x)
+
f (x0 )
= 1
(x0 x1 ) (x0 x1 )(x0 x2 )
(x x0 )(x x1 )
(x x0 )
(x x0 )(x x1 )
+
f (x1 ) +
f (x2 )
+
(x1 x0 ) (x1 x0 )(x1 x2 )
(x2 x0 )(x2 x1 )
(x x0 )(x x2 )
(x x1 )(x x2 )
f (x0 ) +
f (x1 )
=
(x0 x1 )(x0 x2 )
(x1 x0 )(x1 x2 )
(x x0 )(x x1 )
f (x2 )
+
(x2 x0 )(x2 x1 )
:
:
0
3
2
8
4
11
8
18
(x) =
f (x)
0
2
4
8
3
8
11
19
1st divided
dierence
2nd divided
dierence
3rd divided
dierence
5/2
3/2
2
1/4
1/12
1/24
Interpolation
129
(x) = 3 + (x 0)
Thus, it is observed that the interpolating polynomial by both Lagranges and Newtons divided dierence formulae are one and same.
It may be noted that Newtons formula involves less number of arithmetic operations
than that of Lagranges formula.
Algorithm 3.4 (Divided dierence). This algorithm nds the value of y at a
given point x from a table (xi , yi ), i = 0, 1, 2, . . . , n, by Newtons divided dierence
interpolation formula.
Let
yk = f (xk ) = dk,0 .
dk,0 dk1,0
f (xk ) f (xk1 )
=
f [xk , xk1 ] = dk,1 =
xk xk1
xk xk1
dk,1 dk1,1
f [xk , xk1 ] f [xk1 , xk2 ]
f [xk , xk1 , xk2 ] = dk,2 =
=
xk xk2
xk xk2
dk,2 dk1,2
f [xk , xk1 , xk2 , xk3 ] = dk,3 =
xk xk3
In general,
dk,i =
dk,i1 dk1,i1
, i = 1, 2, . . . , n.
xk xki
(3.73)
Using the above notations, the Newtons divided dierence formula (3.68) can be
written in the following form.
f (x) = d0,0 + d1,1 (x x0 ) + d2,2 (x x0 )(x x1 )
+d3,3 (x x0 )(x x1 )(x x2 ) +
+dn,n (x x0 )(x x1 ) (x xn1 )
Interpolation
131
for(i=0;i<=n;i++) printf("%f
%f\n",x[i],y[i]);
for(k=0;k<=n;k++) d[k][0]=y[k];
for(i=1;i<=n;i++)
for(k=i;k<=n;k++)
d[k][i]=(d[k][i1]d[k1][i1])/(x[k]x[ki]);
sum=d[n][n];
for(k=n1;k>=0;k) sum=sum*(xgx[k])+d[k][k];
printf("The interpolated value at x=%f is %f ",xg,sum);
}
A sample of input/output:
Enter number of subintervals 4
Enter x and y values
0.10
1.1052
0.15
1.1618
0.20
1.2214
0.25
1.2840
0.30
1.3499
Enter interpolating point x 0.12
The given values of x and y are
xvalue
yvalue
0.100000
1.105200
0.150000
1.161800
0.200000
1.221400
0.250000
1.284000
0.300000
1.349900
The interpolated value at x=0.120000 is 1.127468
3.16
Inverse Interpolation
In interpolation, for a given set of values of x and y, the value of y is determined for a
given value of x. But the inverse interpolation is the process which nds the value of
x for a given y. Commonly used inverse interpolation formulae are based on successive
iteration.
In the following, three inverse interpolation formulae based on Lagrange, Newton
forward and Newton backward interpolation formulae are described. The inverse interpolation based on Lagranges formula is a direct method while the formulae based on
Newtons interpolation formulae are iterative.
n
i=0
w(x) yi
.
(x xi )w (xi )
n
i=0
w(y)xi
=
Li (y)xi ,
(y yi )w (yi )
n
i=0
where
Li (y) =
w(y)
(yy0 )(yy1 ) (yyi1 )(yyi+1 ) (yyn )
=
.
(yyi )w (yi ) (yi y0 )(yi y1 ) (yi yi1 )(yi yi+1 ) (yi yn )
This formula gives the value of x for given value of y and the formula is known as
Lagranges inverse interpolation formula.
3.16.2
y = y0 + uy0 +
x x0
.
where u =
h
The above formula can be written as
u(u 1) 2
u(u 1)(u 2) 3
1
y0
y0
u =
y y0
y0
2!
3!
u(u 1)(u 2) (u n 1) n
y0 .
n!
Let the rst approximation of u be denoted by u(1) and it is obtained by neglecting
the second and higher dierences as
u(1) =
1
(y y0 ).
y0
Interpolation
133
Next, the second approximation, u(2) , is obtained by neglecting third and higher order
dierences as follows:
u(2) =
1
u(1) (u(1) 1) 2
y y0
y0 .
y0
2!
1
u(2) (u(2) 1) 2
u(2) (u(2) 1)(u(2) 2) 3
y0
y0 .
y y0
y0
2!
3!
In general,
u
(k+1)
1
u(k) (u(k) 1) 2
u(k) (u(k) 1)(u(k) 2) 3
y y0
y0
y0
=
y0
2!
3!
k = 0, 1, 2, . . . .
This process of approximation should be continued till two successive approximations
(k+1)
and u(k) be equal up to desired number of decimal places. Then the value of x is
u
obtained from the relation x = x0 + u(k+1) h.
Example 3.16.1 From the table of values
x
y
:
:
1.8
3.9422
2.0
4.6269
2.2
5.4571
2.4
6.4662
2.6
7.6947
y
3.9422
2.0
4.6269
2 y
3 y
0.6847
0.1455
0.8302
2.2
5.4571
2.4
6.4662
2.6
7.6947
0.0334
0.1789
1.0091
0.0405
0.2194
1.2285
1
1
(5.0 4.6269) = 0.4494.
(y y0 ) =
y0
0.8302
1
u(1) (u(1) 1) 2
u(1) (u(1) 1) 2 y0
y0 = u(1)
y y0
y0
2!
2!
y0
0.4494(0.4494 1) 0.1789
= 0.4761.
0.4494
2
0.8302
u(2) (u(2) 1) 2 y0 u(2) (u(2) 1)(u(2) 2) 3 y0
u(1)
2
y0
3!
y0
0.4761(0.4761 1) 0.1789
0.4494
2
0.8302
0.4761(0.4761 1)(0.4761 2) 0.0405
6
0.8302
0.4494 + 0.0269 0.0031 = 0.4732.
y = yn + vyn +
where v =
That is,
x xn
or x = xn + vh.
h
v =
1
v(v + 1) 2
v(v + 1)(v + 2) 3
yn
yn
y yn
yn
2!
3!
v(v + 1) (v + n 1) n
yn .
n!
Neglecting second and higher order dierences, the rst approximation is given by
v (1) =
Similarly,
v (2) =
1
(y yn ).
yn
1
v (1) (v (1) + 1) 2
y yn
yn .
yn
2!
Interpolation
v (3) =
135
1
v (2) (v (2) + 1) 2
v (2) (v (2) + 1)(v (2) + 2) 3
y yn
yn
yn
yn
2!
3!
and so on.
In general,
v
(k+1)
1
v (k) (v (k) + 1) 2
v (k) (v (k) + 1)(v (k) + 2) 3
yn
yn
y yn
=
yn
2!
3!
k = 0, 1, 2, . . . .
This iteration continues until two consecutive values v (k) and v (k+1) become equal up
to a desired number of signicant gures.
The value of x is given by x = xn + v (k+1) h.
3.16.4
Suppose x = be a root of the equation f (x) = 0 and let it lies between a and b, i.e.,
a < < b. Now, a table is constructed for some values of x, within (a, b), and the
corresponding values of y. Then by inverse interpolation, the value of x is determined
when y = 0. This value of x is the required root.
Example 3.16.2 Find a real root of the equation x3 3x + 1 = 0.
Solution. Let y = x3 3x + 1. One root of this equation lies between 1/4 and 1/2.
Let us consider the points x = 0.25, 0.30, 0.35, 0.40, 0.45, 0.50. The table is shown
below.
x
0.25
0.30
0.35
0.40
0.45
0.50
y
0.265625
0.127000
0.007125
0.136000
0.258875
0.375000
2 y
3 y
0.138625
0.134125
0.128875
0.122875
0.116125
0.00450
0.00525
0.00600
0.00675
0.00075
0.00075
0.00075
1
u(1) (u(1) 1) 2
y0 +
y0
y0
2
1
1.916140 0.916140
0.265625 +
0.00450
0.138625
2
1.944633.
1
u(2) (u(2) 1) 2
u(2) (u(2) 1)(u(2) 2) 3
y0 +
y0 +
y0
y0
2!
3!
1
1.944633 0.944633
0.265625 +
0.00450
0.138625
2
u(1) =
u(2) =
=
=
u(3) =
=
3.17
If the interpolating points are not equally spaced then Lagranges, Newtons divided
dierence or Aitkens iterated interpolation formulae may be used. Newtons forward
formula is appropriate for interpolation at the beginning of the table, Newtons backward formula for interpolation at the end of the table, Stirlings or Bessels formula for
interpolation at the centre of the table. It is well known that the interpolation polynomial is unique and the above formulae are just dierent forms of one and the same
interpolation polynomial and the results obtained by the dierent formulae should be
identical. Practically, only a subset of the set of given interpolating points in the table is used. For interpolation at the beginning of the table, it is better to take this
subset from the beginning of the table. This reason recommends the use of Newtons
forward formula for interpolation at the beginning of the table. For interpolation, near
the end of the table, interpolating points should be available at the end of the table and
hence Newtons backward formula is used for interpolation at the end of the table. For
the same reasons the central dierence formulae like Stirlings, Bessels, Everetts etc.
are used for interpolation near the centre of the table. The proper choice of a central
interpolation formulae depends on the error terms of the dierent formulae.
For interpolation near the centre of the table, Stirlings formula gives the most accurate result for 1/4 s 1/4, and Bessels formula gives most accurate result near
Interpolation
137
s = 1/2, i.e., for 1/4 s 3/4. If all the terms of the formulae are considered, then
both the formulae give identical result. But, if some terms are discarded to evaluate the
polynomial, then Stirlings and Bessels formulae, in general, do not give the same result
and then a choice must be made between them. The choice depends on the order of the
highest dierence that could be neglected so that contributions from it and further differences would be less than half a unit in the last decimal place. If the highest dierence
is of odd order, then Stirlings formula is used and if it is of even order, then, generally,
Bessels formula is used. This conclusion is drawn from the following comparison.
The term of Stirlings formula containing the third dierences is
s(s2 12 ) 3 y1 + 3 y2
.
6
2
This term may be neglected if its magnitude is less than half a unit in the last place,
i.e., if
2
s(s 12 ) 3 y1 + 3 y2 1
.
2
6
2
2
s(s 1)
is 0.064 at s = 1/ 3. Then
The maximum value of
6
3
3
3
3
0.064 y1 + y2 < 1 , i.e., y1 + y2 < 7.8.
2
2
2
The term containing third order dierence of Bessels formula will be less than half
a unit in the last place if
1
s(s 1)(s 1/2) 3
y1 < .
6
2
s(s 1)(s 1/2)
is 0.008 and so that 3 y1  < 62.5.
The maximum value of
6
Thus, if the third dierence is ignored, Bessels formula gives about eight times more
accurate result than Stirlings formula. But, if the third dierences need to be retained
and when the magnitude is more than 62.5, then Everetts formula is more appropriate.
It may be reminded that the Bessels formula with third dierences is equivalent to
Everetts formula with second dierences.
Depending on these discussions the following working rules are recommended for use
of interpolation formulae.
(i) If the interpolating point is at the beginning of the table, then use Newtons
forward formula with a suitable starting point x0 such that 0 < u < 1.
y=4 (x)
=
y=2 (x)
1.5
0 1.5
y=f (x)
 x
Figure 3.4: The graph of the curves y = f (x), y = 2 (x) and y = 4 (x).
For this function y = f (x), the fourth degree polynomial gives an absurd result at
x = 2. At this point f (2) = 0.2, 2 (2) = 0.6 and 4 (2) = 0.01539. It may be
noted that the functions y = f (x) and y = 2 (x) are positive for all values of x, but
y = 4 (x) is negative for some values of x. This example indicates that the higher
degree polynomial does always not give more accurate result.
Interpolation
3.18
139
The interpolation formulae considered so far make use of the function values at some
number of points, say, n + 1 number of points and an nth degree polynomial is obtained.
But, if the values of the function y = f (x) and its rst derivatives are known at n + 1
points then it is possible to determine an interpolating polynomial (x) of degree (2n+1)
which satises the (2n + 2) conditions
(xi ) = f (xi )
(3.74)
(xi ) = f (xi ), i = 0, 1, 2, . . . , n.
This formula is known as Hermites interpolation formula. Here, the number of
conditions is (2n + 2), the number of coecients to be determined is (2n + 2) and the
degree of the polynomial is (2n + 1).
Let us assume the Hermites interpolating polynomial in the form
(x) =
n
hi (x)f (xi ) +
i=0
n
Hi (x)f (xi ),
(3.75)
i=0
where hi (x) and Hi (x), i = 0, 1, 2, . . . , n, are polynomial in x of degree at most (2n + 1).
Using conditions (3.74), we get
1, if i = j
(3.76)
hi (xj ) =
; Hi (xj ) = 0, for all i
0, if i = j
hi (xj ) = 0, for all i; Hi (xj ) =
1, if i = j
0, if i = j
(3.77)
Obviously,
Li (xj ) =
1, if i = j
0, if i = j.
(3.78)
(3.79)
Since each Li (x) is a polynomial of degree n, [Li (x)]2 is a polynomial of degree 2n.
Again, each Li (x) satises (3.79) and [Li (x)]2 also satises (3.79). Since hi (x) and Hi (x)
are polynomials in x of degree (2n + 1), their explicit form may be taken as
hi (x) = (ai x + bi )[Li (x)]2
Hi (x) = (ci x + di )[Li (x)]2
(3.80)
=1
=0
=0
=1
(3.81)
(3.82)
(3.83)
Hence,
n
i=0
n
(3.84)
i=0
:
:
:
1.5
1.14471
0.25438
2.0
1.25992
0.20999
2.5
1.35721
0.18096
Solution.
(x 2.0)(x 2.5)
(x x1 )(x x2 )
=
= 2x2 9x + 10,
(x0 x1 )(x0 x2 )
(1.5 2.0)(1.5 2.5)
(x x0 )(x x2 )
(x 1.5)(x 2.5)
=
= 4x2 + 16x 15,
L1 (x) =
(x1 x0 )(x1 x2 )
(2.0 1.5)(2.0 2.5)
(x 1.5)(x 2.0)
(x x0 )(x x1 )
=
= 2x2 7x + 6.
L2 (x) =
(x2 x0 )(x2 x1 )
(2.5 1.5)(2.5 2.0)
L0 (x) =
Interpolation
141
Therefore
L0 (x) = 4x 9, L1 (x) = 8x + 16, L2 (x) = 4x 7.
Hence L0 (x0 ) = 3, L1 (x1 ) = 0, L2 (x2 ) = 3.
h0 (x) =
=
h1 (x) =
h2 (x) =
=
H0 (x) =
H1 (x) =
H2 (x) =
3
2.8 = 10.07345 0.23040 + 1.25992 2.43360 1.08577 4.32640
+0.33069 0.23040 + 0.16799 2.43360 + 0.05429 4.32640
= 1.40948.
3.19
Spline Interpolation
Spline interpolation is very powerful and widely used method and has many applications
in numerical dierentiation, integration, solution of boundary value problems, two and
three  dimensional graph plotting etc. Spline interpolation method, interpolates a
function between a given set of points by means of piecewise smooth polynomials. In
this interpolation, the curve passes through the given set of points and also its slope
and its curvature are continuous at each point. The splines with dierent degree are
found in literature, among them cubic splines are widely used.
Cubic spline
A function y(x) is called cubic spline in [x0 , xn ] if there exist cubic polynomials
p0 (x), p1 (x), . . . , pn1 (x) such that
y(x) = pi (x) on [xi , xi+1 ], i = 0, 1, 2, . . . , n 1.
pi1 (xi )
pi1 (xi )
=
=
and pi (xi ) = yi ,
pi (xi+1 ) = yi+1 , i = 0, 1, . . . , n 1.
(3.85)
(3.86)
(3.87)
(3.88)
It may be noted that, at the endpoints x0 and xn , no continuity on slope and curvature
are assigned. The conditions at these points are assigned, generally, depending on the
applications.
Let the interval [xi , xi+1 ], i = 0, 1, . . . , n 1 be denoted by ith interval.
Let hi = xi xi1 , i = 1, 2, . . . , n and Mi = y (xi ), i = 0, 1, 2, . . . , n.
Let the cubic spline for the ith interval be
y(x) = ai (x xi )3 + bi (x xi )2 + ci (x xi ) + di ,
in [xi , xi+1 ].
(3.89)
(3.90)
and
yi+1 = ai (xi+1 xi )3 + bi (xi+1 xi )2 + ci (xi+1 xi ) + di
= ai h3i+1 + bi h2i+1 + ci hi+1 + di .
(3.91)
(3.92)
(3.93)
= 6ai hi+1 + 2bi ,
From (3.93), yi = 2bi and yi+1
that is, Mi = 2bi , Mi+1 = 6ai hi+1 + 2bi .
Therefore,
Mi
,
2
Mi+1 Mi
.
ai =
6hi+1
bi =
(3.94)
(3.95)
Interpolation
143
hi+1
6
yi+1 =
(3.96)
(3.97)
(3.98)
hi+1
6
yi yi1 2hi Mi1 + hi Mi
Mi Mi1 2 Mi
.
hi +
= 3
hi +
6hi
2
hi
6
After simplication the above equation reduces to
y
yi yi1
i+1 yi
(3.99)
2M0 + M1 =
and Mn1 + 2Mn
(3.100)
(iv) If M0 = M1
(v) If M0 = y0 and Mn = yn are specied. If a spline satisfy these conditions then it
is called endpoint curvatureadjusted spline.
y
yi yi1
i+1 yi
Let Ai = hi , Bi = 2(hi + hi+1 ), Ci = hi+1 and Di = 6
.
hi+1
hi
Then (3.99) becomes
Ai Mi1 + Bi Mi + Ci Mi+1 = Di , i = 1, 2, . . . , n 1.
(3.101)
0
0
B1 C1 0 0 0
M1
D1
A2 B2 C2 0 0
0
0 M D
2
2
0 A3 B3 C3 0
0
0
(3.102)
.. = ..
. .
Mn1
Dn1
0 0 0 0 0 An1 Bn1
and M0 = Mn = 0.
Imposing the conditions for nonperiodic spline, we nd
2M0 + M1 = D0
and Mn1 + 2Mn = Dn ,
6 y1 y 0
where D0 =
y0
h1
h1
6
yn yn1
and Dn =
y
.
hn n
hn
(3.103)
(3.104)
(3.105)
Interpolation
145
Then equations (3.101), (3.103), (3.104) and (3.105) result the following tridiagonal
system for the unknowns M0 , M1 , . . . , Mn .
M D
0
0
2 1 0 0 0
0
0
D1
M
A1 B1 C1 0 0
1
0
0
D2
M2
0 A2 B2 C2 0
0
0
. = .
.
(3.106)
. .
. .
h1 (M2 M1 )
hn (Mn1 Mn2 )
and Mn = Mn1 +
.
h2
hn1
(3.107)
A1 h1
A1 h1
+ M2 C1
= D1 or, M1 B1 + M2 C1 = D1
M1 A1 + B1 +
h2
h2
A1 h1
A1 h1
and C1 = C1
.
h2
h2
= Dn1
Similarly, the second expression is transferred to Mn2 An1 + Mn1 Bn1
where
Cn1 hn
hn Cn1
An1 = An1
and Bn1
= Bn1 + Cn1 +
.
hn1
hn1
For this case, the tridiagonal system of equations for M1 , M2 , . . ., Mn1 is
0
0
B1 C1 0 0 0
M1
D1
A2 B2 C2 0 0
0
0
M2
D2
0 A3 B3 C3 0
0
0
=
.
(3.108)
.
.
. .
.
.
Mn1
Dn1
0 0 0 0 0 An1 Bn1
where B1 = A1 + B1 +
D1
0
0
B1 C1 0 0 0
M1
D2
A2 B2 C2 0 0
0
0
M2
.. ,
. =
0 A3 B3 C3 0
0
0
(3.109)
. .
.
Dn2
Mn1
0 0 0 0 0 An1 Bn1
D
n1
where
D1
= D1
A1 y0 ,
Dn1
= Dn1
Cn1 yn .
Example 3.19.1 Fit a cubic spline curve that passes through (0, 0.0), (1, 0.5), (2,
2.0) and (3, 1.5) with the natural end boundary conditions, y (0) = y (3) = 0.
Solution. Here the intervals are (0, 1) (1, 2) and (2, 3), i.e., three intervals of
x, in each of which we can construct a cubic spline. These piecewise cubic spline
polynomials together gives the cubic spline curve y(x) in the entire interval (0, 3).
Here h1 = h2 = h3 = 1.
Then equation (3.99) becomes
Mi1 + 4Mi + Mi+1 = 6(yi+1 2yi + yi1 ),
i = 1, 2, 3.
That is,
M0 + 4M1 + M2 = 6(y2 2y1 + y0 ) = 6 (2.0 2 0.5 + 0.0) = 6
M1 + 4M2 + M3 = 6(y3 2y2 + y1 ) = 6 (1.5 2 2.0 + 0.5) = 12.
Imposing the conditions M0 = y (0) = 0 and M3 = y (3) = 0 to the above equations,
and they simplify as
4M1 + M2 = 6,
M1 + 4M2 = 12.
12
18
and M2 = .
These equations give M1 =
5
5
Let the natural cubic spline be given by
pi (x) = ai (x xi )3 + bi (x xi )2 + ci (x xi ) + di
where the coecients ai , bi , ci and di are given by the relations
Mi+1 Mi
Mi
,
,
bi =
6hi+1
2
yi+1 yi 2hi+1 Mi + hi+1 Mi+1
,
ci =
hi+1
6
ai =
d i = yi ,
for i = 0, 1, 2.
Therefore,
M1 M 0
= 0.4,
6
y1 y0 2M0 + M1
= 0.1,
c0 =
1
6
M2 M 1
= 1,
a1 =
6
a0 =
b0 =
M0
= 0,
2
d0 = y0 = 0.
b1 =
6
M1
= ,
2
5
Interpolation
y2 y1 2M1 + M2
= 1.3,
1
6
3
M3 M 2
= ,
a2 =
6
5
y3 y2 2M2 + M3
= 0.7,
c2 =
1
6
147
d1 = y1 = 0.5.
c1 =
b2 =
9
M2
= ,
2
5
d2 = y2 = 2.0.
Hence the required piecewise cubic splines in each interval are given by
p0 (x) = 0.4x3 + 0.1x,
0x1
3
p1 (x) = (x 1) + 1.2(x 1)2 + 1.3(x 1) + 0.5,
p2 (x) = 0.6(x 2)3 1.8(x 2)2 + 0.7(x 2) + 2.0,
1x2
2 x 3.
Example 3.19.2 Fit a cubic spline curve for the following data with end conditions
y (0) = 0.2 and y (3) = 1.
x
y
:
:
0
0
1
0.5
2
3.5
3
5
Solution. Here, the three intervals (0, 1) (1, 2) and (2, 3) are given in each of which
the cubic splines are to be constructed. These cubic spline functions are denoted by
y0 , y1 and y2 . In this example, h1 = h2 = h3 = 1.
For the boundary conditions, equation (3.99) is used. That is,
M0 + 4M1 + M2 = 6(y2 2y1 + y0 )
M1 + 4M2 + M3 = 6(y3 2y2 + y1 ).
Also, from the equations (3.100)
2M0 + M1 = 6(y1 y0 y0 ) and M2 + 2M3 = 6(y3 y3 + y2 )
i.e.,
M0 + 4M1 + M2
M1 + 4M2 + M3
2M0 + M1
M2 + 2M3
=
=
=
=
15
9
1.8
6(1 5 + 3.5) = 15.
ci =
hi+1
6
ai =
di = yi , for i = 0, 1, 2.
Therefore,
M1 M 0
= 0.98, b0 = 0.68, c0 = 0.2, d0 = 0.
6
M2 M 1
= 1.04, b1 = 2.26, c1 = 1.78, d1 = 0.5.
a1 =
6
M3 M 2
= 0.82, b2 = 0.86, c2 = 3.18, d2 = 3.5.
a2 =
6
a0 =
Hence, the required piecewise cubic spline polynomials in each interval are given by
y0 (x) = 0.98x3 0.68x2 + 0.2x,
0x1
3
2
y1 (x) = 1.04(x 1) + 2.26(x 1) + 1.78(x 1) + 0.5,
y2 (x) = 0.82(x 2)3 0.86(x 2)2 + 3.18(x 2) + 3.5,
1x2
2 x 3.
2 x + 26x 2 x + 18, 1 x 2,
f (x) =
1 x 2,
2 x 3.
pi (xi ) = f (xi ),
pi1 (xi ) = pi (xi ),
pi (xi+1 ) = f (xi+1 ),
pi1 (xi ) = pi (xi ),
i = 0, 1 and
i = 1.
Interpolation
149
But, here the values of f (x0 ), f (x1 ) and f (x2 ) are not supplied, so only the conditions
of (b) are to be checked.
Now,
33
75
33
189
p1 (x) = x2 80x +
p0 (x) = x2 + 52x ,
2
2
2
2
p1 (x) = 33x 80.
p0 (x) = 33x + 52,
p0 (x1 ) = p0 (2) = 0.5, p1 (x1 ) = p1 (2) = 0.5, i.e., p0 (x1 ) = p1 (x1 ).
p0 (x1 ) = p0 (2) = 14 and p1 (x1 ) = p1 (2) = 14. Thus p0 (x1 ) = p1 (x1 ).
Hence f (x) is a spline.
Algorithm 3.5 (Cubic spline). This algorithm nds the cubic spline for each of
the intervals [xi , xi+1 ], i = 0, 1, . . . , n 1. The (i) Natural spline, (ii) Nonperiodic
spline or clamped cubic spline, (iii) Extrapolated spline, and (iv) endpoint curvatureadjusted spline are incorporated here. The spacing for xi need not be equal and
assume that x0 < x1 < < xn .
Algorithm Cubic Spline
Read xi , yi , i = 0, 1, 2, . . . , n;//xs are not necessary equispaced//
Read xg; //the value of x at which y is to be computed.//
//Computation of hi .//
for i = 1 to n do
hi = xi xi1 ;
endfor;
//Compututation of Ai , Bi , Ci and Di //
for i = 1 to n 1 do
Bi = 2(hi + hi+1
);
Ci = hi+1 ;
Ai = hi;
yi+1 yi yi yi1
;
Di = 6
hi+1
hi
endfor;
Case :
(i) Natural spline
To nd M1 , M2 , . . . , Mn1 , solve the system of tridiagonal equation
dened in (3.102).
Set M0 = Mn = 0.
(ii) Nonperiodic spline
y at x = x0 ,
xn //
Read y0 , yn ; //rstderivative of
6 y1 y 0
6
yn yn1
Compute D0 =
y
.
y0 , D n =
h1
h1
hn n
hn
To nd M0 , M1 , M2 , . . . , Mn , solve the system of tridiagonal equation
dened in (3.106).
hi+1
6
endfor;
//Printing of splines//
for i = 0 to n 1 do
Print Coecient of , i, th spline is ai , bi , ci , di ;
endfor;
//Computation of y at x = xg//
if (xg < x0 ) or (xg > xn ) then
Print x outside the range;
Stop;
endif;
for i = 0 to n 1 do
if (xg < xi+1 ) then
j = i;
exit from for loop;
endif;
endfor;
Compute yc = aj (xg xj )3 + bj (xg xj )2 + cj (xg xj ) + dj ;
Print The value of y at x =, xg, is, yc ;
end Cubic Spline
Interpolation
Program 3.5
.
/* Program Cubic Spline
This program construct cubic splines at each interval
[x[i1],x[i]], i=1, 2, ..., n and finds the value of
y=f(x) at a given x when the function is supplied as
(x[i],y[i]), i=0, 1, ..., n. */
#include<stdio.h>
#include<math.h>
#include<ctype.h>
#include<stdlib.h>
float M[21];
void main()
{
int i,n;
char opt,s[5];
float x[20],y[20],h[20],A[20],B[20],C[20],D[20];
float a[20],b[20],c[20],d[20],xg,yd0,ydn,temp,yc;
float TriDiag(float a[],float b[],float c[],float d[],int n);
printf("\nEnter number of subintervals ");
scanf("%d",&n);
printf("Enter x and y values ");
for(i=0;i<=n;i++) scanf("%f %f",&x[i],&y[i]);
printf("Enter interpolating point x ");
scanf("%f",&xg);
printf("The given values of x and y are\nxvalue
yvalue\n");
for(i=0;i<=n;i++) printf("%f
%f\n",x[i],y[i]);
for(i=0;i<=n;i++) h[i]=x[i]x[i1]; /* computation of h[i] */
for(i=1;i<n;i++) /* computation of A,B,C,Ds */
{
A[i]=h[i];
B[i]=2*(h[i]+h[i+1]);
C[i]=h[i+1];
D[i]=6*((y[i+1]y[i])/h[i+1](y[i]y[i1])/h[i]);
}
printf("\nN Natural spline\n");
printf("P NonPeriodic spline\n");
printf("E Extrapolated spline\n");
printf("C End point Curvature adjusted spline\n");
151
printf("
Enter your choice ");
opt=getche();
switch(toupper(opt))
{
case N: /* Natural spline */
temp=TriDiag(A,B,C,D,n1);
M[0]=0; M[n]=0;
break;
case P: /* Nonperiodic spline */
printf("\nEnter the values of y[0] and y[n] ");
scanf("%f %f",&yd0,&ydn);
D[0]=6*((y[1]y[0])/h[1]yd0)/h[1];
D[n]=6*(ydn(y[n]y[n1])/h[n])/h[n];
for(i=n+1;i>=1;i) D[i]=D[i1];
A[n+1]=1; B[n+1]=2;
for(i=n;i>=2;i){
A[i]=A[i1]; B[i]=B[i1]; C[i]=C[i1];}
B[1]=2; C[1]=1;
temp=TriDiag(A,B,C,D,n+1);
for(i=0;i<=n;i++) M[i]=M[i+1];
break;
case E: /* Extrapolated spline */
B[1]=A[1]+B[1]+A[1]*h[1]/h[2];
C[1]=C[1]A[1]*h[1]/h[2];
A[n1]=A[n1]C[n1]*h[n]/h[n1];
B[n1]=B[n1]+C[n1]+C[n1]*h[n]/h[n1];
temp=TriDiag(A,B,C,D,n1);
M[0]=M[1]h[1]*(M[2]M[1])/h[2];
M[n]=M[n1]+h[n]*(M[n1]M[n2])/h[n1];
break;
case C: /* End point Curvature adjusted spline */
printf("\nEnter the values of y[0] and y[n] ");
scanf("%f %f",&yd0,&ydn);
D[1]=D[1]A[1]*yd0;
D[n1]=D[n1]C[n1]*ydn;
temp=TriDiag(A,B,C,D,n1);
M[0]=yd0; M[n]=ydn;
break;
Interpolation
153
Interpolation
N
P
E
C
Natural spline
NonPeriodic spline
Extrapolated spline
End point Curvature adjusted spline
Enter your choice n
Natural spline
NonPeriodic spline
Extrapolated spline
End point Curvature adjusted spline
Enter your choice p
Enter the values of y[0] and y[n] 0 1
The cubic splines are
p0(x)= 0.6250(x+ 1.0000)^3 1.3750(x+ 1.0000)^2+ 0.0000(x+ 1.0000)
+ 1.0000 in [1.0000, 1.0000]
155
3.20
Bivariate Interpolation
Like single valued interpolation, recently bivariate interpolations become important due
to their extensive uses in a wide range of elds e.g., digital image processing, digital
lter design, computeraided design, solution of nonlinear simultaneous equations etc.
In this section, some of the important methods are described to construct interpolation formulae that can be eciently evaluated.
To construct the formulae, the following two approaches are followed.
(i) Constructing a function that matches exactly the functional values at all the data
points.
(ii) Constructing a function that approximately ts the data. This approach is desirable when the data likely to have errors and require smooth functions.
On the basis of these approaches, one can use four types of methods (i) local matching
methods, (ii) local approximation methods, (iii) global matching methods and (iv) global
approximation methods. In the local methods, the constructed function at any point
depends only on the data at relatively nearby points. In global methods, the constructed
function at any points depends on all or most of the data points.
In the matching method, the matching function matches exactly the given values,
but in the approximate method the function approximately ts the data.
Here, the local and the global matching methods are discussed only.
3.20.1
Here two local matching methods are described, viz., triangular interpolation and rectangular grid or bilinear interpolation.
Triangular interpolation
The simplest local interpolating surface is of the form
F (x, y) = a + bx + cy.
The data at the three corners of a triangle determine the coecients. This procedure
generates a piecewise linear surface which is global continuous.
Interpolation
157
Suppose the function f (x, y) be known at the points (x1 , y1 ), (x2 , y2 ) and (x3 , y3 ).
Let f1 = f (x1 , y1 ), f2 = f (x2 , y2 ) and f3 = f (x3 , y3 ).
Let the constructed function be
F (x, y) = a + bx + cy
(3.110)
a =
where
= (x2 x1 )(y3 y1 ) (x3 x1 )(y2 y1 ).
The values of a, b, c give the required polynomial.
But, the function F (x, y) can be written in the following form
F (x, y) = Af1 + Bf2 + Cf3 ,
(x2 x)(y3 y) (x3 x)(y2 y)
where A =
Note 3.20.1
(3.111)
(3.112)
(3.113)
(3.114)
(i) If A + B + C = 1 then = 0.
Example 3.20.1 For a function f (x, y), let f (1, 1) = 8, f (2, 1) = 12 and f (2, 2) =
20. Find the approximate value of f (3/2, 5/4) using triangular interpolation.
Solution. Here given that
x1 = 1,
x2 = 2,
x3 = 2,
3
x= ,
2
y1 = 1,
y2 = 1,
y3 = 2,
5
y= .
4
f1 = f (x1 , y1 ) = 8
f2 = f (x2 , y2 ) = 12
f3 = f (x3 , y3 ) = 20
Therefore,
= (x2 x1 )(y3 y1 ) (x3 x1 )(y2 y1 )
= (2 1)(2 1) (2 1)(1 1) = 1.
1
(x2 x)(y3 y) (x3 x)(y2 y)
=
A =
2
1
(x3 x)(y1 y) (x1 x)(y3 y)
=
B =
4
1
(x1 x)(y2 y) (x2 x)(y1 y)
= .
C =
4
Thus f (3/2, 5/4) F (3/2, 5/4) = Af1 + Bf2 + Cf3 =
1
2
8+
1
4
12 +
1
4
20 = 12.
Bilinear interpolation
Let a function f (x, y) be known at the points (x1 , y1 ), (x1 + h, y1 ), (x1 , y1 + k) and
(x1 + h, y1 + k).
A function F (x, y) is to be constructed within the rectangle formed by these points.
Let f1 = f (x1 , y1 ), f2 = f (x1 + h, y1 ), f3 = f (x1 , y1 + k) and
f4 = f (x1 + h, y1 + k).
Let us construct a function F (x, y) of the form
F (x, y) = a + b(x x1 ) + c(y y1 ) + d(x x1 )(y y1 )
(3.115)
such that
F (x1 , y1 ) = f (x1 , y1 ) = f1 ,
F (x1 + h, y1 ) = f (x1 + h, y1 ) = f2 ,
Interpolation
Thus
a = f1 ,
c =
f2 f1
,
h
f4 + f1 f2 f3
and d =
.
hk
159
b=
f3 f1
k
(3.116)
Example 3.20.2 Find a bilinear interpolation polynomial F (x, y) for the function
f (x, y) where f (1, 1) = 8, f (2, 1) = 10, f (1, 2) = 12 and f (2, 2) = 20. Also, nd an
approximate value of f (4/3, 5/3).
Solution. Here
y1 = 1,
f1 = f (x1 , y1 ) = 8
x1 = 1,
y1 = 1,
f2 = f (x1 + h, y1 ) = 10
x1 + h = 2,
y1 + k = 2,
f3 = f (x1 , y1 + k) = 12
x1 = 1,
y1 + k = 2,
f4 = f (x1 + h, y1 + k) = 20.
x1 + h = 2,
Obviously, h = 1, k = 1.
10 8
f2 f1
Thus,
=
= 2,
a = f1 = 8, b =
h
1
12 8
f4 + f1 f2 f3
f3 f1
=
= 4, d =
= 6.
c=
k
1
hk
Hence,
f (x, y) F (x, y) = a + b(x x1 ) + c(y y1 ) + d(x x1 )(y y1 )
= 8 + 2(x 1) + 4(y 1) + 6(x 1)(y 1).
38
Therefore,
f (4/3, 5/3) = .
3
3.20.2
Global methods
n
n
aij xi y j .
(3.117)
i=1 j=1
F (x, y) = [1 y y 2
y n1 ] .
..
..
.. ..
.
.
.
.
. .
xn1
a1n a2n ann
(3.118)
F (xn , y1 ) F (xn , y2 ) F (xn , yn )
1 x1 x21 x1n1
1 y1 y12 y1n1
n1
2
1 x2 x2 xn1
2
2
, Yt (yj ) = 1 y2 y2 y2 .
X(xi ) =
2
n1
1 xn xn xn
1 yn yn2 ynn1
Since the matrices X, Y and F are known, one can calculate the matrix A as (assuming X and Y are nonsingular)
A = (Y1 )t FX1 .
(3.119)
6
10
10
18
Yt (y) = (1 y),
n = 2.
1 1
1 1
t
X(xi ) =
, Y (yj ) =
,
1 2
1 2
1
6 10
1 1
A =
10 18
1 2
2 1
6 10
2 1
2 0
=
=
.
1 1
10 18
1 1
0 4
1 1
1 2
Interpolation
Therefore,
2 0
F (x, y) = [1 y]
0 4
1
x
161
= 2 + 4xy.
i = 0, 1, . . . , m; j = 0, 1, . . . , n.
(3.120)
i = 0, 1, . . . , m
(3.121)
j = 0, 1, . . . , n
(3.122)
m
n
(3.124)
i=0 j=0
1
1.732051
1.414214
2
1
1
i=0 j=0
Therefore,
F (x, y) = (1 x)(1 y) 1 + (1 x)y 1.414214 + x(1 y) 1.732051 + 2xy
= 1 + 0.732051x + 0.414214y 0.146265xy.
Thus,
F (0.25, 0.75) = 1 + 0.732051 0.25 + 0.414214 0.75 0.146265 0.25 0.75
= 1.466248563.
Interpolation
wy
wx
fij ;
(xg xi )wdx(i) (yg yj )wdy(j)
endfor;
endfor;
Print The value of f (x, y) is , sum;
end Lagrange Bivariate
function wdx(j)
sum = 0;
for i = 0 to m do
if (i = j) sum = sum + (xj xi );
endfor;
return sum;
end wdx(j)
function wdy(j)
sum = 0;
for i = 0 to n do
if (i = j) sum = sum + (yj yi );
endfor;
return sum;
end wdy(j)
Program 3.6
.
/* Program Lagrange bivariate
This program is used to find the value of a function
f(x,y) at a given point (x,y) when a set of values of
f(x,y) is given for different values of x and y, by
Lagrange bivariate interpolation formula. */
#include<stdio.h>
#include<math.h>
float x[20],y[20];
void main()
{
int i,j,n,m;
float xg,yg,f[20][20],wx=1,wy=1,sum=0;
float wdx(int j,int m); float wdy(int j,int n);
printf("Enter the number of subdivisions along x and y ");
scanf("%d %d",&m,&n);
printf("Enter x values ");
for(i=0;i<=m;i++) scanf("%f",&x[i]);
163
Interpolation
f(1.000000,2.000000)= 9
f(2.000000,0.000000)= 6
f(2.000000,1.000000)= 9
f(2.000000,2.000000)= 14
Enter the interpolating point 0.5 0.5
The interpolated value at ( 0.50000, 0.50000) is
165
2.75000
+mnxy + f (x0 , y0 ).
Substituting m =
y y0
x x0
and n =
.
h
k
x x0
y y0
x +
y f (x0 , y0 )
F (x, y) = f (x0 , y0 ) +
h
k
2(x x0 )(y y0 )
1 (x x0 )(x x1 )
xy
xx +
+
2!
h2
hk
(y y0 )(y y1 )
+
yy f (x0 , y0 ) +
k2
Then m 1 =
Thus
(3.125)
(3.126)
Example 3.20.5 For the following data obtain Newtons bivariate interpolating
polynomial and hence calculate the values of f (0.75, 0.25) and f (1.25, 1.5).
y
x
0
1
2
1
1
5
3
2
1
5
5
3
Solution.
x f (x0 , y0 ) =
=
y f (x0 , y0 ) =
=
xx f (x0 , y0 ) =
=
yy f (x0 , y0 ) =
=
xy f (x0 , y0 ) =
=
f (x0 + h, y0 ) f (x0 , y0 )
f (x1 , y0 ) f (x0 , y0 ) = 1 1 = 2
f (x0 , y0 + k) f (x0 , y0 )
f (x0 , y1 ) f (x0 , y0 ) = 3 1 = 2
f (x0 + 2h, y0 ) 2f (x0 + h, y0 ) + f (x0 , y0 )
f (x2 , y0 ) 2f (x1 , y0 ) + f (x0 , y0 ) = 5 2 (1) + 1 = 2
f (x0 , y0 + 2k) 2f (x0 , y0 + k) + f (x0 , y0 )
f (x0 , y2 ) 2f (x0 , y1 ) + f (x0 , y0 ) = 5 2 3 + 1 = 0
f (x0 + h, y0 + k) f (x0 , y0 + k) f (x0 + h, y0 ) + f (x0 , y0 )
f (x1 , y1 ) f (x0 , y1 ) f (x1 , y0 ) + f (x0 , y0 ) = 1.
Interpolation
Here h = k = 1. u =
Thus,
167
y y0
x x0
= x, v =
= y.
h
k
F (x, y) = 1 + [x (2) + y 2]
1
+ [x(x 1) (2) + 2xy 1 + y(y 1) 0]
2!
= 1 x + 2y x2 + xy.
Hence f (0.75, 0.25) F (0.75, 0.25) = 0.375 and f (1.25, 1.5) F (1.25, 1.5) = 3.0625.
3.21
x2 + 2x + 3
(x + 1)x(x 1)
:
:
1
2
0
3
1
6
f (x) =
Hence
x2 + 2x + 3
f (x)
1
3
3
=
=
+
.
(x + 1)x(x 1)
(x + 1)x(x 1)
x+1 x x1
:
:
0
1
1
2
2
4
3
?
4
16
Solution.
Method 1.
Using Lagranges formula
(x 1)(x 2)(x 4)
(0 1)(0 2)(0 4)
(x 0)(x 2)(x 4)
L1 (x) =
(1 0)(1 2)(1 4)
(x 0)(x 1)(x 4)
L2 (x) =
(2 0)(2 1)(2 4)
(x 0)(x 1)(x 2)
L3 (x) =
(4 0)(4 1)(4 2)
L0 (x) =
Therefore,
x3 7x2 + 14x 8
.
8
x3 6x2 + 8x
=
.
3
x3 5x2 + 4x
=
.
4
x3 3x2 + 2x
=
.
24
=
1
5
11
, c = and d = .
12
8
24
Interpolation
169
Therefore,
y(x) = 1 +
1
11
5
x x2 + x3 .
12
8
24
:
:
2.0
0.693147
2.1
0.741937
(x) =
2 2.1.
1
The maximum value of f (x) = 2 in 2 x 2.1 is f (2.0) = 0.25.
x
Then
0.25
E1 (x) (2.05 2)(2.05 2.1)
= 0.000313.
2
Thus the upper bound of truncation error is 0.000313.
Example 3.21.4 For the following table nd the value of y at x = 2.5, using
piecewise linear interpolation.
x
y
:
:
1
35
2
40
3
65
4
72
5
80
2.5 2
2.5 3
40 +
65 = 52.5.
23
32
Example 3.21.5 Deduce the following interpolation formula taking three points
x0 , x0 + , 0 and x1 using Lagranges formula.
f (x) =
where
1
E(x) = (x x0 )2 (x x1 )f () and min{x0 , x0 + , x1 } max{x0 , x0 + , x1 }.
6
Solution. The Lagranges interpolating polynomial for the points x0 , x0 + and x1
is
(x x0 )(x x1 )
(x x0 )(x x1 )
f (x0 ) +
f (x0 + )
f (x)
(x0 x0 )(x0 x1 )
(x0 + x0 )(x0 + x1 )
(x x0 )(x x0 )
f (x1 ) + E(x)
+
(x1 x0 )(x1 x0 )
(x x0 )(x x1 )
(x x0 )(x x1 )
f (x0 ) +
f (x0 )
=
(x0 x1 )
(x0 + x1 )
(x x0 )(x x0 )
f (x1 ) + E(x)
+
(x1 x0 )(x1 x0 )
(x x0 )(x x0 )
f (x0 ) + E(x)
+
(x1 x0 )(x1 x0 )
(x x0 )(x1 x)
(x1 x)(x + x1 2x0 )
f (x0 ) +
f (x0 )
=
2
(x1 x0 )
(x1 x0 )
(x x0 )2
+
f (x1 ) + E(x) as 0.
(x1 x0 )2
The error term is
E(x) = (x x0 )(x x0 )(x x1 )
=
f ()
3!
1
(x x0 )2 (x x1 )f (), as 0
6
Interpolation
171
P (x)
0.682689
0.706282
0.728668
0.749856
0.769861
0.788700
2 P
3 P
0.023593
0.022386
0.021188
0.020005
0.018839
0.001207
0.001198 0.000009
0.001183 0.000015
0.001166 0.000017
Example 3.21.7 Find the seventh and the general terms of the series 3, 9, 20, 38,
65, . . ..
Solution. Let xi = i, i = 1, 2, 3, 4, 5 and y0 = 3, y1 = 9, y2 = 20, y3 = 38, y4 = 65.
We construct the Newtons backward interpolation polynomial using these values.
x
1
2
3
4
5
y
3
9
20
38
65
y 2 y 3 y
6
11
18
27
5
7
9
2
2
(x) = yn + vyn +
=
=
=
=
The seventh term is
1
(7) = (2 73 + 3 72 + 13 7) = 154.
6
[Other interpolation formulae may also be used to solve this problem.]
Example 3.21.8 From the following table of sin x compute sin 120 and sin 450 .
x
: 100
200
300
400
500
y = sin x : 0.17365 0.34202 0.50000 0.64279 0.76604
Solution. The dierence table is
x
y
0
10 0.17365
2 y
3 y
4 y
0.16837
0.01039
200 0.34202
0.15798
300
0.01519
0.50000
0.14279
0.01954
400 0.64279
0.00480
0.00045
0.00435
0.12325
500 0.76604
(i) To nd sin 120 .
Here x0 = 100 , x = 120 , h = 100 , u =
By Newtons forward formula
(xx0 )
h
(120 100 )
100
= 0.2.
u(u 1) 2
u(u 1)(u 2) 3
y0 +
y0
2!
3!
Interpolation
2 u x i
0.056
0.106
0.122
0.135
3 u x i
0.050
0.016
0.013
173
u32 = y0 + s
x is tabulated below.
x : 5600
5700
5800
5900
6000
y : 17.75808 17.86316 17.96702 18.06969 18.17121
Compute
2 y
3 y
4 y
0.10508
1 5700 17.86316
0.00122
0.10386
0.00003
0.00119
0 5800 17.96702
0.10267
0.00001
0.00004
0.00115
1 5900 18.06969
0.10152
2 6000 18.17121
(i) For x = 5860, let us take x0 = 5800, then s = (5860 5800)/100 = 0.6.
By Bessels formula
s(s 1) 2 y0 + 2 y1
y 0 + y1
+ (s 0.5)y0 +
2
2!
2
1
3
+ (s 0.5)s(s 1) y1
3!
17.96702 + 18.06969
+ (0.6 0.5) 0.10267
=
2
y(5860) =
Interpolation
y1 + y0 s2 2
s(s2 1) 3 y2 + 3 y1
+ y1 +
2
2!
3!
2
0.10386 + 0.10267 0.62
= 17.96702 + 0.6
+
(0.00119)
2
2
0.6(0.62 1) 0.00003 + 0.00004
+
6
2
= 18.02877.
y(5860) = y0 + s
Thus
5860 = 18.02877.
Example 3.21.11 Prove that the third order divided dierence of the function
1
1
.
f (x) = with arguments a, b, c, d is
x
abcd
1
.
x
1
1
1
f (b) f (a)
= b a = .
f [a, b] =
ba
ba
ab
1 + 1
1
f [a, b] f [b, c]
= ab bc =
.
ac
ac
abc
The third order divided dierence is
f [a, b, c] =
f [a, b, c] f [b, c, d]
f [a, b, c, d] =
=
ad
1
bcd
1
=
.
ad
abcd
1
abc
1
(1)n
, prove that f [x0 , x1 , . . . , xn ] =
.
x
x0 x1 xn
f (x1 ) f (x0 )
1
(1)1
= x1 x0 =
=
.
x1 x0
x1 x0
x0 x1
x0 x1
1
f [x0 , x1 ] =
175
1 + 1
f [x0 , x1 ] f [x1 , x2 ])
(1)2
= x0 x1 x1 x2 =
.
x0 x2
x0 x2
x0 x1 x2
(1)k
.
x0 x1 xk
(1)k
1
(1)k
=
x0 xk+1 x0 x1 xk
x1 x2 xk+1
k
1
1
(1)
1
=
x1 x2 xk x0 xk+1 x0 xk+1
(1)k+1
.
=
x0 x1 x2 xk+1
f [x0 , x1 , . . . , xk , xk+1 ] =
f (x0 ) f (x1 )
u(x0 )v(x0 ) u(x1 )v(x1 )
=
x0 x1
x0 x1
u(x0 )[v(x0 ) v(x1 )] + v(x1 )[u(x0 ) u(x1 )]
=
x0 x1
= u(x0 )v[x0 , x1 ] + v(x1 )u[x0 , x1 ].
f [x0 , x1 ] =
Interpolation
177
Example 3.21.14 Show that the nth divided dierence f [x0 , x1 , . . . , xn ] can be
expressed as
f [x0 , x1 , . . . , xn ] = D V, where
and
1
1
1
x0
x
x
1
2
2
2
2
x0
x
x
1
2
D =
n1
x
xn1
xn1
1
2
0
y0
y1
y2
1
xn
x2n
xnn1
yn
1
1
1
x0
x
x
1
2
2
2
2
x0
x
x
1
2
V =
n1 n1 n1
x
x
x
1
2
0n
x
xn1
xn2
0
1
xn
x2n
xnn1
xnn
.
1
x0
V (x0 , x1 , . . . , xn1 , x) = x20
n
x0
1
x1
x21
xn1
1
xn
x2n
xn1
n
xnn
.
1
xn1 x
x2n1 x2 .
xnn1 xn
1
n1
(xn xi ).
i=0
j=0
i>j
1
1 1
1 1
x0
x
x
x
1
i1
i+1 xn
2
2
2
2
2
x
x
x
x
C(x0 , x1 , . . . , xi1 , xi+1 , . . . , xn ) = 0
1
i1
i+1 xn
....................................
n1 n1
x
x
xn1 xn1 xnn1
0
i1
i+1
.
Interpolation
179
Now,
D = (1)n y0 C(x1 , x2 , . . . , xn ) + (1)n+1 y1 C(x0 , x2 , . . . , xn ) +
+(1)n+n yn C(x0 , x1 , . . . , xn1 )
n
(1)i yi V (x0 , x1 , . . . , xi1 , xi+1 , . . . , xn ).
=
i=0
Thus,
DV =
n
i=0
n
(1)i yi
i=0
n
i=0
(1)i
(xi x0 )(xi x1 ) (xi xi1 )(xi xi+1 ) (xi xn )
yi
(xi x0 )(xi x1 ) (xi xi1 )(xi xi+1 ) (xi xn )
= f [x0 , x1 , . . . , xn ].
(y) =
1
(137 + 700 300) = 16.78125
32
1
and x(5) (5) = (137 + 350 75) = 12.87500.
32
Hence,
x(10) (10) =
Therefore,
x (0) =
4
=
=
=
=
=
0.08249
= 0.02234.
3.69194
0.31054
= 0.47684.
0.65125
0.21997
= 0.88176.
0.24947
0.09961
= 0.63966.
0.15577
0.08249
= 0.30338.
0.27190
i=0
Interpolation
3.22
181
Exercise
1. Show that
n
i=1
w(x)
= 1.
(x xi )w (xi )
when f (x) =
9. From the following table, nd the number of students who obtain less than 35
marks
Marks
: 2030 3040 4050 5060 6070
No. of Students : 32
53
50
38
33
10. If y(1) = 3, y(3) = 9, y(4) = 30 and y(6) = 132, nd the Lagranges interpolation
polynomial that takes the same values as the function y at the given points.
11. Let the following observation follows the law of a cubic polynomial
4
5
6
13
18
24
16. The following table gives pressure of a steam plant at a given temperature. Using
Newtons formula, compute the pressure for a temperature of 1420 C.
Temperature 0 C : 140 150 160 170
180
2
Pressure, kgf/cm : 3.685 4.854 6.302 8.076 10.225
Interpolation
183
17. The following data gives the melting point of an alloy of lead and zinc; where T
is the temperature in 0 C and P is the percentage of lead in the alloy. Find the
melting point of the alloy containing 84% of lead using Newtons interpolation
method.
P : 50 60 70 80
T : 222 246 272 299
18. Using a polynomial of third degree, complete the record of the export of a certain
commodity during ve years, as given below.
Year, x
: 1988 1989 1990 1991 1992
Export in tons, y : 450 388 400 470
19. Find the polynomial which attains the following values at the given points.
x : 1 0 1 2 3 4
f (x) : 16 7 4 1 8 29
20. Compute log10 2.5 using Newtons forward dierence interpolation formula, given
that
x
:
2.0
2.2
2.4
2.6
2.8
3.0
log10 x : 0.30103 0.34242 0.38021 0.41497 0.44716 0.47721
21. Find the missing term in the following table:
x : 0 1 2 3 4
y : 1 3 9 81
Why the result diers from 33 = 27 ?
22. In the following table, the value of y are consecutive terms of a series of which the
number 36 is the fth term. Find the rst and the tenth terms of the series. Find
also the polynomial which approximates these values.
x : 3 4 5 6 7 8 9
y : 18 26 36 48 62 78 96
23. From the following table determine (a) f (0.27), and (b) f (0.33).
x : 0.24
0.26
0.28
0.30
0.32
0.34
f (x) : 1.6804 1.6912 1.7024 1.7139 1.7233 1.7532
26. Using Gausss backward formula, nd the value of 518 given that
0
:
0
5
10
15
20
log E : 0.196120 0.195293 0.192815 0.188690 0.182928
Compute log 120 by (a) Bessels formula and (b) Stirlings formula and compare
the results.
30. The value of the elliptic integral
%
E() =
/2
(1 sin2 )1/2 d
for certain equidistance values of are given below. Use Everetts or Bessels
formula to determine E(0.25).
Interpolation
185
:
0.20
0.22
0.24
0.26
0.28
0.30
E() : 1.659624 1.669850 1.680373 1.691208 1.702374 1.713889
31. Using Everetts formula, evaluate f (20) from the following table.
x : 14
18
22
26
f (x) : 2877 3162 3566 3990
32. Using Aitkens method evaluate y when x = 2 from the following table.
x : 1 3 4 6
y : 3 9 30 132
33. Use the Aitkens procedure to determine the value of f (0.2) as accurately as
possible from the following table.
x : 0.17520 0.25386 0.33565 0.42078 0.50946
f (x) : 0.84147 0.86742 0.89121 0.91276 0.93204
34. Show that the rst order divided dierence of a linear polynomial is independent
of the arguments.
35. Show that the second order divided dierence of a quadratic polynomial is constant.
36. If f (x) is continuous for x0 x x1 , show that f [x0 , x1 ] = f () where x0
x1 and hence show that
f [x0 , x0 ] lim f [x0 , x1 ] = f (x0 ).
x1 x0
37. For the equidistant values x0 , x1 , x2 , x3 i.e., xi = x0 + ih, establish the following
relations
1
[f (x1 ) f (x0 )],
h
1
[f (x2 ) 2f (x1 ) + f (x0 )],
f [x0 , x1 , x2 ] =
2!h2
1
[f (x3 ) 3f (x2 ) + 3f (x1 ) f (x0 )].
and f [x0 , x1 , x2 , x3 ] =
3!h3
f [x0 , x1 ] =
38. If f (x) =
ax + b
obtain expressions for f [p, q], f [p, p, q] and f [p, p, q, q].
cx + d
1
.
(a x0 )(a x1 ) (a xn )
Interpolation
187
46. Use the technique of inverse interpolation to nd x for which sinh x = 5.5 from
the following table.
x : 2.2
2.4
2.6
2.8
sinh x : 4.457 5.466 6.095 8.198
47. Given the following table of f (x) between x = 1.1 and x = 1.5, nd the zero of
f (x).
x : 1.1
1.2
1.3
1.4
1.5
f (x) : 1.769 1.472 1.103 1.344 1.875
48. Use the technique of inverse interpolation to nd a real root of the equation
x3 2x 4 = 0.
49. Using Hermites interpolation formula, estimate the value of log 3.2 from the following table
x
:
3.0
3.5
4.0
y = log x : 1.09861 1.25276 1.38629
: 0.33333 0.28571 0.25000
y = x1
50. Find the Hermite polynomial of the third degree approximating the function y(x)
such that
y(x0 ) = 1, y(x1 ) = 0 and y (x0 ) = y (x1 ) = 0.
51. The following values of x and y are calculated from the relation y = x3 + 10
x : 1 2 3 4 5
y : 11 18 37 74 135
Determine the cubic spline p(x) for the interval [2, 3] given that
(a) p (1) = y (1) and p (5) = y (5), (b) p (1) = y (1) and p (5) = y (5).
52. Fit a cubic spline to the function dened by the set of points given in the following
table.
x
: 0.10
0.15
0.20
0.25
0.30
x
y = e : 1.1052 1.1618 1.2214 1.2840 1.3499
Chapter 4
4.1
Location of Roots
An interval [a, b] is said to be the location of a real root c if f (c) = 0 for a < c < b.
Mainly, two methods are used to locate the real roots of an equation, one is graphical
method and other is an analytic method known as method of tabulation.
4.1.1
Graphical method
First method:
In this method, the graph of y = f (x) is drawn in a rectangular coordinates system.
It is obvious that the abscissas of the points where the graph intersects the xaxis are
the roots of the equation f (x) = 0. But, practically, it is most dicult to determine
the exact value of x where the graph intersects the xaxis. For example, if x = 1.27831
is a root of an equation f (x) = 0 then we can not determine 1.2783 (four digits after
decimal point) from the graph. We can measure the value of x up to one or two decimal
places. But, the approximate value of the root can be determined using this method.
Second method:
Some times, the approximate roots of f (x) = 0 can be determined by dividing all
the terms of the equation into groups, one of them is written on the lefthand side of
the equation and the other on the right hand side, i.e., the equation is represented as
g(x) = h(x). Then the graph of two functions y = g(x) and y = h(x), are drawn. The
abscissas of the points of intersection of these graphs are the roots of the equation.
Example 4.1.1 Use graphical method to locate the roots of the equation x3 4x
2 = 0.
Solution. First method:
The graph of the function y = x3 4x 2 is drawn in Figure 4.1(a). The curve cuts
the xaxis at three points and, consequently, the equation has three real roots. From
gure, it is observed that the roots belong to the intervals [2, 1], [1, 0] and [2, 3].
Second method:
The given equation can be written as x3 = 4x + 2. The graph of the functions y = x3
and y = 4x + 2 are drawn (Figure 4.1(b)). The abscissas of the points of intersection
of the graphs of these functions are roots of the equations. The intervals of the roots
are [2, 1], [1, 0] and [2, 3].
yM = 4x + 2
191
yI
= x3
O
2
1
 x
Method of tabulation
This method depends on the continuity of the function f (x). Before applying the
tabulation method, following result should be noted.
Theorem 4.1 If f (x) is continuous in the interval (a, b) and if f (a) and f (b) have the
opposite signs, then at least one real root of the equation f (x) = 0 lies within the interval
(a, b).
If f (a) and f (b) have same signs then f (x) = 0 has no real roots or has an even
number of real roots.
If the curve y = f (x) touches the xaxis at some point, say, x = c then c is a root
of f (x) = 0 though f (a) and f (b), a < c < b may have same sign. For example,
f (x) = (x 2)2 touches the xaxis at x = 2, also f (1.5) > 0 and f (2.5) > 0, but, x = 2
is the root of the equation f (x) = (x 2)2 = 0.
A trial method for tabulation is as follows:
Form a table of signs of f (x) setting x = 0, 1, 2, . . .. If the sign f (x) changes it signs
for two consecutive values of x then at least one root lies between these two values, i.e.,
if f (a) and f (b) have opposite signs then a root lies between a and b.
Example 4.1.2 Find the location of roots of the equation 8x3 20x2 2x + 5 = 0
by tabulation method.
Example 4.1.3 Find the number of real roots of the equation 3x 3x 2 = 0 and
locate them.
Solution. Let f (x) = 3x 3x 2. The domain of denition of the function f (x) is
(, ).
Now, f (x) = 3x log 3 3.
The roots of f (x) = 0 is given by 3x log 3 3 = 0
log 3 log log 3
3
or, x =
= 0.914.
log 3
log 3
A table of signs of f (x) is then form by setting x equal to
(a) the values close of the roots of f (x) = 0, i.e., x = 0, x = 1 and
(b) boundary values of domain, i.e., x = .
or, 3x =
x
0 1
Sign of f (x) + +
The equation 3x 3x 2 = 0 has two real roots since the function twice changes sign,
among them one is negative root and other is greater than 1.
193
A new table with small intervals of the location of the root is constructed in the
following.
x
0 1 1 2
Sign of f (x) + +
The roots of the given equation are in (1, 0) [as f (0).f (1) < 0] and (1, 2).
This section is devoted to locate the roots which is the rst stage of solution of
algebraic and transcendental equations.
The second stage is the computation of roots with the specied degree of accuracy. In
the following sections some methods are discussed to determine the roots of an algebraic
or a transcendental equation. Before presenting the solution methods we dene the order
of convergence of a sequence of numbers in the following.
Order of Convergence
Assume that the sequence {xn } of numbers converges to and let n = xn for n 0.
If two positive constants A = 0 and p > 0 exist and
lim
n+1
=A
pn
(4.1)
then the sequence is said to converge to with order of convergence p. The number A
is called the asymptotic error constant.
If p = 1, the order of convergence of {xn } is called linear and if p = 2, the order of
convergence is called quadratic, etc.
In the next section, one of the bracketing method called bisection method is introduced.
4.2
Bisection Method
Let be a root of the equation f (x) = 0 lies in the interval [a, b], i.e., f (a).f (b) < 0,
and (b a) is not suciently small. The interval [a, b] is divided into two equal intervals
a+b
ba
, and c =
(Figure 4.2). If f (c) = 0, then c is
[a, c] and [c, b], each of length
2
2
an exact root.
Now, if f (c) = 0, then the root lies either in the interval [a, c] or in the interval [c, b].
If f (a).f (c) < 0 then the interval [a, c] is taken as new interval, otherwise [c, b] is taken
as the next interval. Let the new interval be [a1 , b1 ] and use the same process to select
the next new interval. In the next step, let the new interval be [a2 , b2 ]. The process of
bisection is continued until either the midpoint of the interval is a root, or the length
(bn an ) of the interval [an , bn ] (at nth step) is suciently small. The number an and
an + bn
is taken
2
a
O
 x
(4.2)
(4.3)
.
(4.4)
log 2
For example, if the length of the interval is b a = 1 and = 0.0001, then n is given
by n 14.
The minimum number of iterations required to achieved the accuracy for b a = 1
are shown in Table 4.1.
Theorem 4.2 Assume that f (x) is a continuous function on [a, b] and that there exists
a number [a, b] such that f () = 0. If f (a) and f (b) have opposite signs, and {xn }
represents the sequence of midpoints generated by the bisection method, then
 xn 
ba
for n = 0, 1, 2, . . .
2n+1
(4.5)
195
Proof. The root and the midpoint xn both lie in the interval [an , bn ], the distance
between xn and cannot be greater than half the width of the interval [an , bn ]. Thus
 xn 
bn an 
for all n.
2
(4.6)
From the bisection method, it is observed that the successive interval widths form
the following pattern.
b0 a0 
, where b0 = b and a0 = a,
21
b0 a0 
b1 a1 
=
,
b2 a2  =
2
22
b2 a2 
b0 a0 
.
=
b3 a3  =
2
23
b1 a1  =
b0 a0 
.
2n
 xn 
b0 a0 
[using (4.6)].
2n+1
i.e., lim xn = .
n
Note 4.2.1 If the function f (x) is continuous on [a, b] then the bisection method is
applicable. This is justied in Figure 4.3. For the function f (x) of the graph of Figure
4.3, f (a) f (b) < 0, but the equation f (x) = 0 has no root between a and b as the
function is not continuous at x = c.
b
O
 x
Figure 4.3: The function has no root between a and b, though f (a) f (b) < 0.
Note 4.2.2 This method is very slow, but it is very simple and will converge surely
to the exact root. So the method is applicable for any function only if the function is
continuous within the interval [a, b], where the root lies.
In this method derivative of the function f (x) and premanipulation of function are
not required.
Note 4.2.3 This method is also called bracketing method since the method successively
reduces the two endpoints (brackets) of the interval containing the real root.
Example 4.2.1 Find a root of the equation x2 + x 7 = 0 by bisection method,
correct up to two decimal places.
Solution. Let f (x) = x2 + x 7.
f (2) = 1 < 0 and f (3) = 5 > 0. So, a root lies between 2 and 3.
n
0
1
2
3
4
5
6
7
8
9
10
f (xn+1 )
1.750
0.313
0.359
0.027
0.143
0.062
0.018
0.003
0.008
0.002
0.002
197
Algorithm 4.1 (Bisection method). This algorithm nds a real root of the equation f (x) = 0 which lies in [a, b] by bisection method.
Algorithm Bisection
Input function f (x);
// Assume that f (x) is continuous within [a, b] and a root lies on [a, b].//
Read ;
//tolerance for width of the interval//
Read a, b;
//input of the interval//
Compute f a = f (a); f b = f (b);
//compute the function values//
if sign(f a) = sign(f b) then
//sign(f a) gives the sign of the value of f a.//
Print f (a) f (b) > 0, so there is no guarantee for a root within [a, b];
Stop;
endif;
do
Compute c = (a + b)/2;
Compute f c = f (c);
if f c = 0 or f c < then
a = c and b = c;
else if sign(f b) = sign(f c) then
b = c; f b = f c;
else
a = c; f a = f c;
endif;
while (b a > );
Print the desired root is c;
end Bisection
Program 4.1
.
/* Program Bisection
Program to find a root of the equation x*x*x2x1=0 by
bisection method.
Assume that a root lies between a and b. */
#include<stdio.h>
#include<math.h>
#include<stdlib.h>
#define f(x) x*x*x2*x1 /* definition of the function f(x) */
void main()
{
float a,b,fa,fb,c,fc;
float eps=1e5; /* error tolerance */
4.3
The RegulaFalsi method is one of the most widely used methods of solving algebraic
and transcendental equations. This method is also known as method of false position,
199
(4.7)
To nd the point of intersection, set y = 0 in (4.7) and let (x2 , 0) be such point.
Thus,
x2 = x0
f (x0 )
(x1 x0 ).
f (x1 ) f (x0 )
(4.8)
This is the second approximation of the root. Now, if f (x2 ) and f (x0 ) are of opposite
signs then the root lies between x0 and x2 and then we replace x1 by x2 in (4.8). The
next approximation is obtained as
x3 = x0
f (x0 )
(x2 x0 ).
f (x2 ) f (x0 )
If f (x2 ) and f (x1 ) are of opposite signs then the root lies between x1 and x2 and the
new approximation x3 is obtain as
x3 = x2
f (x2 )
(x1 x2 ).
f (x1 ) f (x2 )
The procedure is repeated till the root is obtained to the desired accuracy.
If the nth approximate root (xn ) lies between an and bn then the next approximate
root is thus obtained as
xn+1 = an
f (an )
(bn an ).
f (bn ) f (an )
(4.9)
f (x)
x0 =a
O
x1 x2 s x
b
x0
x1x2
 x
(4.10)
where is an exact root and xn1 and xn are its approximations obtained at the (n1)th
and nth iterations. This relation can be used when
M 2m, where M = max f (x) and m = min f (x) in [a, b].
(4.11)
left end
point an
0.0000
0.6700
0.7570
0.7690
0.7707
right end
point bn
1.0
1.0
1.0
1.0
1.0
f (an )
2.0000
0.3600
0.0520
0.0072
0.0010
f (bn )
1.0
1.0
1.0
1.0
1.0
xn+1
0.6700
0.7570
0.7690
0.7707
0.7709
f (xn+1 )
0.3600
0.0520
0.0072
0.0010
0.0001
201
do
{
x2=(x0*fx1x1*fx0)/(fx1fx0);
fx2=f(x2);
if(fabs(fx2)<eps)
{
printf("The root is %8.5f ",x2);
exit(0);
}
if(fx2*fx0<0)
{
x1=x2; fx1=fx2;
}
else
{
x0=x2; fx0=fx2;
}
}while(fabs(fx2)>eps);
} /* main */
A sample of input/output:
Enter the value of x0 and x1 0 3
The root is 3.00000
4.4
(4.12)
Suppose x0 [a, b] be an initial guess to the desired root . Then (x0 ) is evaluated
and this value is denoted by x1 . It is the rst approximation of the root . Again, x1 is
substituted for x to the right side of (4.12) and obtained a new value x2 = (x1 ). This
process is continued to generate the sequence of numbers x0 , x1 , x2 , . . . , xn , . . ., those
are dened by the following relation:
xn+1 = (xn ),
n = 0, 1, 2, . . .
(4.13)
203
This successive iterations are repeated till the approximate numbers xn s converges
to the root with desired accuracy, i.e., xn+1 xn  < , where is a suciently small
number. The function (x) is called the iteration function.
Note 4.4.1 There is no guarantee that this sequence x0 , x1 , x2 , . . . will converge. The
function f (x) = 0 can be written as x = (x) in many dierent ways. This is very
signicant since the form of the function (x) is very important both for the convergence
and for its rate.
For example, the equation x3 + x2 1 = 0 has a root lies between 0 and 1. This
equation can be rewritten in the following ways:
1
1 x2
1 x2
2 1/3
, etc.
;
x
=
;
x
=
(1
x
)
;
x
=
1 x3 ; x =
x=
2
x
x
1+x
The following theorem gives the sucient condition for convergence of the iteration
process.
Theorem 4.3 Let be a root of the equation f (x) = 0 and it can be written as x = (x)
and further that
1. the function (x) is dened and dierentiable on the interval [a, b],
2. (x) [a, b] for all x [a, b],
3. there is a number l < 1 such that
 (x) l < 1 for x [a, b].
(4.14)
Then the sequence {xn } given by (4.13) converges to the desired root irrespective of
the choice of the initial approximation x0 [a, b] and the root is unique.
Proof. Since is a root of the equation x = (x), therefore
= ().
(4.15)
xi+1 = (xi ).
(4.16)
(4.17)
Hence the sequence {xn } converge to if  (x) < 1, for all x [a, b].
Now to prove the uniqueness.
Let 1 and 2 be two roots of x = (x), i.e., 1 = (1 ) and 2 = (2 ). Then
2 1  = (2 ) (1 ) =  (c)1 2 ,
(4.18)
where c (1 , 2 ).
Equation (4.18) reduces to
1 2 (1  (c)) = 0
and by condition (iii) 1 = 2 , i.e., the two roots are not distinct, they are equal.
4.4.1
Estimation of error
l
ln
xn xn1 
x1 x0 .
1l
1l
(4.19)
205
Let the maximum number of iterations needed to achieve the accuracy be N ().
Thus from (4.19)
lN
x1 x0  .
1l
This gives
log x(1l)
1 x0 
.
(4.20)
N ()
log l
For l
1
2
(4.21)
Order of convergence
The convergence of an iteration method depends on the suitable choice of the iteration
function (x) and x0 , the initial guess.
Let xn converges to the exact root , so that = ().
Thus xn+1 = (xn ) ().
Let n+1 = xn+1 . Note that (x) = 0. Then the above relation becomes
n+1 = (n + ) ()
1
= n () + 2n () +
2
= n () + O(2n )
i.e.,
n+1 n ().
f (x)
y=x
y=x
?
6 66
6
?
??
?
6
y = (x)
?
y = (x)
 x
 x
x0x2 x3x1
x1
x0
x2
O
O
(b) Spiral case solution,
(a) Stair case solution,
1 < () < 0.
0 < () < 1
Figure 4.5: Convergent for  () < 1.
f (x)
6
y=(x)
f (x)
y=x
6
y=x
6
6
x0
x1
x
O
x3x1
x0x2 x4
y = (x)

Example 4.4.1 Consider the equation 5x3 20x + 3 = 0. Find the root lying on
the interval [0,1] with an accuracy of 104 .
3
xn
(xn ) = xn+1
0.5
0.18125
0.18125
0.15149
0.15149
0.15087
0.15087
0.15086
0.15086
0.15086
207
At this stage the iteration process is terminated and = 0.1509 is taken as the
required root.
Example 4.4.2 Find a root of the equation
cos x xex = 0
correct up to three decimal places.
Solution. It is easy to see that one root of the given equation lies between 0 and 1.
Let x0 = 0. The equation can be written as x = ex cos x = (x) (say).
The calculations are shown in the following table.
n
0
1
2
3
4
5
6
xn 0.50000 0.53228 0.50602 0.52734 0.51000 0.52408 0.51263
xn+1 0.53228 0.50602 0.52734 0.51000 0.52408 0.51263 0.52193
n
7
8
9
10
11
12
13
xn 0.52193 0.51437 0.52051 0.51552 0.51958 0.51628 0.51896
xn+1 0.51437 0.52051 0.51552 0.51958 0.51628 0.51896 0.51678
n
14
15
16
17
18
19
20
xn 0.51678 0.51855 0.51711 0.51828 0.51733 0.51810 0.51748
xn+1 0.51855 0.51711 0.51828 0.51733 0.51810 0.51748 0.51798
Therefore, the required root is 0.518 correct up to three decimal places.
Algorithm 4.3 (Fixed point iteration). This algorithm computes a root of the
equation f (x) = 0 by rewriting the equation as x = (x), provided  (x) < 1 in the
interval [a, b], by xed point iteration method. x0 [a, b] be the initial guess and
is the error tolerance.
Algorithm Iteration
Input function (x);
Read x0 , ; //initial guess and error tolerance.//
Set x1 = x0 ;
do
Set x0 = x1 ;
Compute x1 = (x0 );
while (x1 x0  > );
Print The root is, x1 ;
end Iteration
Program 4.3
.
/* Program FixedPoint Iteration
Program to find a root of the equation x*x*x3x+1=0
by fixed point iteration method. phi(x) is obtained
by rewrite f(x)=0 as x=phi(x), which is to be supplied.*/
#include<stdio.h>
#include<math.h>
#include<stdlib.h>
#define phi(x) (3*x1)/(x*x)
/* definition of the function phi(x) and it to be
changed accordingly */
void main()
{
int k=0; /* counts number of iterations */
float x1,x0; /* initial guess */
float eps=1e5; /* error tolerance */
printf("\nEnter the initial guess x0 ");
scanf("%f",&x0);
x1=x0;
do
{
k++;
x0=x1;
x1=phi(x0);
}while(fabs(x1x0)>eps);
printf("One root is %8.5f obtained at %d th iteration ",x1,k);
} /* main */
A sample of input/output:
Enter the initial guess x0 1
One root is 1.53209 obtained at 37th iteration
4.5
The rate of convergence of iteration method is linear. But, this slow rate can be accelerated by using Aitkens method.
The iteration scheme of this method is obtained from xed point iteration method as
xn+1 = (xn ) with  (x) < 1.
209
(xn+1 xn )2
.
xn+1 2xn + xn1
(4.22)
(xn )2
,
2 xn1
(4.23)
2 x0
0.58101
The results for n = 1, 2, 3, 4 are shown below.
Then x3 = x2
n
1
2
3
4
xn1
0.00000
0.54030
0.49958
0.50244
xn
0.54030
0.49959
0.50244
0.50225
xn+1
0.49959
0.50244
0.50225
0.50226
xn
0.04071
0.00285
0.00019
0.00001
2 xn1
0.58101
0.04356
0.00304
0.00021
xn+2
0.50244
0.50225
0.50226
0.50226
4.6
h2
f (x0 ) + = 0.
2!
Neglecting the second and higher order derivatives the above equation reduces to
f (x0 ) + hf (x0 ) = 0 or, h =
f (x0 )
.
f (x0 )
Hence,
x1 = x0 + h = x0
f (x0 )
.
f (x0 )
(4.24)
To compute the value of h, the second and higher powers of h are neglected so the
f (x0 )
is not exact, it is an approximate value. So, x1 , obtained from
value of h =
f (x0 )
(4.24) is not a root of the equation, but it is a better approximation of x than x0 .
In general,
xn+1 = xn
f (xn )
.
f (xn )
(4.25)
211
f (x)
6
6
6
6
x2
x1
x0
 x
z0
O
x0
y0
 x
and
x0 = 2.
Then x1 = 4.0, x2 = 5.33333, . . . , x15 = 19.72255, . . . and clearly {xn } diverges slowly
to (Figure 4.9).
f (x)
6
U
6
 x
R6s
Figure 4.9: NewtonRaphson method produces a divergent sequence for f (x) = xex .
Now consider a function f (x) = x3 x 3 which will produce a cyclic sequence when
initial guess is x0 = 0. The sequence is
x1 = 3.0, x2 = 1.961538, x3 = 1.147176, x4 = 0.006579,
x5 = 3.000389, x6 = 1.961818, x7 = 1.147430, . . .
and it may be noted that xk+4 xk , k = 0, 1, 2, . . . (Figure 4.10).
But, the initial guess x0 = 2 gives the convergent sequence x1 = 1.72727, x2 =
1.67369, x3 = 1.67170, x4 = 1.67170.
The function f (x) = tan1 x and x0 = 1.45 gives a divergent oscillating sequence. If
the initial guess is x0 = 1.45 then
x1 = 1.55026, x2 = 1.84593, x3 = 2.88911, . . .
(Figure 4.11). But, if x0 = 1.5 then x1 = 0.07956, x2 = 0.00034, x3 = 0.00000.
x1
x2
x3
?
x0
213
 x
y = x3 x 3
?
f (xn )
.
f (xn )
Comparing this expression with xed point iteration formula xn+1 = (xn ) and we
obtain
f (xn )
.
(xn ) = xn
f (xn )
This can be written as
(x) = x
f (x)
.
f (x)
It is already proved that the iteration method converges if  (x) < 1. Therefore,
NewtonRaphson method converges, if
d
f
(x)
(4.26)
or
f (x) f (x) < f (x)2
dx x f (x) < 1
within the interval under consideration. NewtonRaphson method converges if the initial
guess x0 is chosen suciently close to the root and the functions f (x), f (x) and f (x)
are continuous and bounded in any small interval containing the root. The rate of
convergent of NewtonRaphson method is stated in the following theorem.
y = tan1 x
66
x )
x3
x0
x1
x2
 x
6
?
?
f (xn )
.
f (xn )
Let xn = n + .
Therefore, above relation becomes
f (n + )
f (n + )
f () + n f () + (2n /2)f () +
n
[by Taylors series]
f () + n f () +
2n f ()
f () n + 2 f () +
[as f () = 0]
n
f ()
f () 1 + n f () +
2n f ()
f ()
+ 1 n
+
n n +
2 f ()
f ()
f ()
2 f ()
+ 2n
+ O(3n )
n
2 f ()
f ()
1 2 f ()
+ O(3n ).
2 n f ()
n+1 + = n +
n+1 =
=
=
=
215
Neglecting the terms of order 3n and higher powers the above expression becomes
n+1 = A2n , where A =
f ()
.
2f ()
(4.27)
This relation shows that NewtonRaphson method has quadratic convergence or second order convergence.
Example 4.6.1 Use NewtonRaphson method to nd a root of the equation x3 +
x 1 = 0.
Solution. Let f (x) = x3 + x 1. Then f (0) = 1 < 0 and f (1) = 1 > 0. So one
root lies between 0 and 1. Let x0 = 0 be the initial root.
The iteration scheme is
f (xn )
f (xn )
2x3 + 1
x3 + xn 1
= n2
.
= xn n 2
3xn + 1
3xn + 1
xn+1 = xn
xn
0
1
0.7500
0.6861
0.6823
xn+1
1
0.7500
0.6861
0.6823
0.6823
xn+1 = xn
or, xn+1
Example 4.6.3 Write down an iteration scheme for nding square root of a positive
number N . Hence nd the square root of the number 2.
Second Part. Let x = 2 or, x2 2 = 0. Also, let f (x) = x2 2. Then f (x) = 2x.
The NewtonRaphson iteration scheme is
x2 2
x2 + 2
= n
.
xn+1 = xn n
2 xn
2 xn
Let x0 = 1. The successive calculations are shown in the following.
n
0
1
2
3
Therefore, the value of
xn
1
1.50000
1.41667
1.41422
xn+1
1.5000
1.41667
1.41422
1.41421
217
3 x2n + 2
is an iteration scheme to
8
nd a root of the equation f (x) = 0. Find the function f (x).
3 x2n + 2
.
8
Therefore, lim xn = .
n
1
Thus, lim xn+1 = 3 lim x2n + 2 .
n
8 n
1
This gives = [32 + 2], i.e., 32 + 2 = 8 or, 32 8 + 2 = 0.
8
Thus the required equation is 3x2 8x + 2 = 0 and hence f (x) = 3x2 8x + 2.
Example 4.6.6 Discuss the NewtonRaphson method to nd the root of the equation x10 1 = 0 starting with x0 = 0.5.
Solution. The real roots of this equation are 1.
Here f (x) = x10 1.
Therefore,
x10 1
9x10 + 1
xn+1 = xn n 9 = n 9 .
10xn
10xn
9 (0.5)10 + 1
= 51.65, which is far away from the root 1.
10 (0.5)9
This is because 0.5 was not close enough to the root x = 1.
But the sequence {xn } will converge to the root 1, although very slowly.
The initial root x0 = 0.9 gives the rst approximate root x1 = 1.068, which is close
to the root 1.
This example points out the role of initial approximation in NewtonRaphson method.
When x0 = 0.5 then x1 =
f (zn )
.
f (zn )
n
0
1
2
3
4
5
6
zn
(0.50000, 0.50000)
(1.00000, 1.00000)
(0.75000, 0.75000)
(0.41781, 0.71918)
(0.55230, 0.85744)
(0.49763, 0.86214)
(0.50003, 0.86603)
f (zn )
( 0.25000, 0.25000)
( 1.00000,0.00000)
( 0.34375, 0.09375)
( 0.05444, 0.24110)
( 0.08475,0.02512)
(0.00014, 0.00785)
( 0.00004,0.00003)
f (zn )
(0.00000, 0.50000)
(2.00000,2.00000)
(1.00000,0.37500)
(0.69919, 1.07384)
(1.49971, 0.58836)
(1.47746, 0.87439)
(1.50005, 0.86587)
zn+1
(1.00000, 1.00000)
(0.75000, 0.75000)
(0.41781, 0.71918)
(0.55230, 0.85744)
(0.49763, 0.86214)
(0.50003, 0.86603)
(0.50000, 0.86603)
4.7
f (xn )
f (xn )
(4.28)
gives a faster convergent sequence. The term p1 f (xn ) is the slope of the straight line
passing through (xn , f (xn )) and intersecting the xaxis at the point (xn+1 , 0). The
formula (4.28) reduces to NewtonRaphson formula when p = 1.
If is a root of f (x) = 0 with multiplicity p, then is also a root of f (x) = 0
with multiplicity (p 1), of f (x) = 0 with multiplicity (p 2) and so on. Hence the
expression
x0 p
f (x0 )
,
f (x0 )
x0 (p 1)
f (x0 )
,
f (x0 )
x0 (p 2)
f (x0 )
,...
f (x0 )
should have the same value if there is a root with multiplicity p, when the initial guess
is very close to the exact root .
Theorem 4.5 The rate of convergence of the formula (4.28) is quadratic.
Proof. Let be a multiple root of multiplicity p, of the equation f (x) = 0. Then
f () = f () = f () = = f p1 () = 0 and f p () = 0. Let n = xn . Then from
(4.28),
n+1 = n p
= n p
f (n + )
f (n + )
f () + n f () + +
f () + n f () + +
p1
p1 ()
n
(p1)! f
p2
n
(p2)! f
p1 ()
pn p
p! f ()
p1
n
(p1)! f
p+1
p+1 ()
n
(p+1)! f
p ()
pn
p!
f p+1 () +
= n p
=
=
=
=
pn p
p! f ()
p1
n
p+1
p+1 ()
n
(p+1)! f
219
pn
f p+1 () +
1
2n
f p+1 ()
n f p+1 ()
n
+
+ 1 +
+
n p
p
p(p + 1) f p ()
p f p ()
2n
f p+1 ()
n f p+1 ()
n
+
+ 1
+
n p
p
p(p + 1) f p ()
p f p ()
2 f p+1 () 2n f p+1 ()
n n + n
+
p + 1 f p ()
p f p ()
f p+1 ()
1
2
+ O(3n ).
n
p(p + 1) f p ()
(p1)! f
p ()
p!
f p+1 ()
1
.
p(p + 1) f p ()
This shows that the rate of convergence is quadratic.
Thus n+1 = A2n , where A =
4.8
In the NewtonRaphson method, the derivative of the function f (x) is calculated at each
point xn . That is, at each iteration two functions are evaluated at xn , n = 0, 1, 2, . . ..
f (xn )
.
f (x0 )
(4.29)
That is, the derivative of f (x) is calculated only at the initial guess instead of several
dierent points xn . This method reduces the time for calculating the derivatives. But,
the rate of convergence of this method is linear, which is proved in Theorem 4.6.
Theorem 4.6 The rate of convergence of the formula (4.29) is linear.
Solution. Let be the root of the equation f (x) = 0. Then f () = 0, and n = xn .
Therefore from (4.29),
f (n + )
f () + n f () +
n+1 = n
=
n
f (x0 )
f (x0 )
f ()
+ O(2n ).
= n 1
f (x0 )
Neglecting 2n and higher powers of 2n and denoting A = 1
f ()
the above error
f (x0 )
term becomes
n+1 = An .
(4.30)
This proved that the rate of convergence of the formula (4.29) is linear.
f (x)
6
6
s
6/
x3x2 x1
x0
221
Geometrical interpretation
The gradient of tangent at the point xn is f (x0 ) for all n.
Thus the line passing through the point (xn , f (xn )) is parallel to the tangent drawn
at (x0 , f (x0 )), i.e., the tangent at (xn , f (xn )) in NewtonRaphson method is replaced
by a line parallel to the tangent drawn at (x0 , f (x0 )) and passing through the point
(xn , f (xn )). This phenomena is shown in Figure 4.12.
Example 4.8.1 Find a root of the equation x3 x + 1 = 0 using formula (4.29)
and NewtonRaphson method up to four decimal places.
Solution. One root of this equation lies between 2 and 1. Let x0 = 1.5 and
f (x) = x3 x + 1. Then f (x0 ) = 5.75.
The iteration scheme of the formula (4.29) is
f (xn )
f (x0 )
x3 xn + 1
1
=
(x3 6.75xn + 1).
= xn n
5.75
5.75 n
xn+1 = xn
xn
1.50000
1.34783
1.33032
1.32614
1.32508
1.32481
1.32474
1.32472
xn+1
1.34783
1.33032
1.32614
1.32508
1.32481
1.32474
1.32472
1.32472
Therefore, one root of the given equation is 1.3247, correct up to four decimal places
attained at 7th iteration.
Using NewtonRaphson method
The iteration scheme for NewtonRaphson method is
f (xn )
f (xn )
2x3 1
x3 xn + 1
= n2
.
= xn n 2
3xn 1
3xn 1
xn+1 = xn
xn
1.50000
1.34783
1.32520
1.32472
xn+1
1.34783
1.32520
1.32472
1.32472
Therefore, a root is 1.3247 correct up to four decimal places, attained at 3rd iteration.
This example shows that NewtonRaphson method is more faster than the method
given by (4.29).
4.9
f (xn
f (xn )
= (xn ) (say)
+ a(xn )f (xn ))
(4.31)
That is,
(x) = x
f (x
f (x)
,
+ a(x)f (x))
(4.32)
f (x)
f (x + a(x)f (x))
f (x)f (x + a(x)f (x))(1 + a (x)f (x) + a(x)f (x))
+ 2
{f (x + a(x)f (x))}2
f (x){f (x + a(x)f (x))}2 {1 + a (x)f (x) + a(x)f (x)}2
2
{f (x + a(x)f (x))}3
and (x) =
1
H.H.H.Homeier, A modied Newton method for root nding with cubic convergence, J. Computational and Applied Mathematics, 157 (2003) 227230.
223
() =
(4.33)
1
1
then () = 0. This is the easiest way to put a(x) =
.
2f ()
2f (x)
Hence the iteration scheme for modied NewtonRaphson method is
Now, if a() =
xn+1 = xn
f (xn )
1
, where a(xn ) =
.
f (xn + a(xn )f (xn ))
2f (xn )
(4.34)
Also we have
() = , () = 0, () = 0.
(4.35)
1
().
3!
(4.36)
xn
1.50000
1.85202
1.95215
1.98420
1.99468
1.99803
1.99916
f (xn )
0.62500
0.06245
0.00676
0.00074
0.00008
0.00001
0.000002
g(xn )
1.63889
1.89000
1.96421
1.98816
1.99601
1.99852
1.99937
f (g(xn ))
1.77546
0.62370
0.21090
0.07062
0.02389
0.00887
0.00378
xn+1
1.85202
1.95215
1.98420
1.99468
1.99803
1.99916
1.99969
Therefor, a root of the given equation is 2.000, correct up to three decimal places,
and this value is attained at 6th iteration, while NewtonRaphson method takes 10
iterations (see Example 4.7.1).
Algorithm 4.4 (NewtonRaphson method). This algorithm nds a root of the
equation f (x) = 0 by NewtonRaphson method, when f (x), f (x) and initial guess
x0 are supplied.
Algorithm NewtonRaphson
// f d(x) is the derivative of f (x) and is the error tolerance.//
Input function f (x), f d(x);
Read x0 , ;
Set x1 = x0 ;
do
Set x0 = x1 ;
Compute x1 = x0 f (x0 )/f d(x0 );
while (x1 x0  > );
Print The root is, x1 ;
end NewtonRaphson
225
Program 4.4
.
/* Program NewtonRaphson
Program to find a root of the equation x*x*x3x+1=0 by NewtonRaphson method. f(x) and its derivative fd(x) are to be supplied. */
#include<stdio.h>
#include<math.h>
#include<stdlib.h>
void main()
{
int k=0; /* counts number of iterations */
float x1,x0; /* x0 is the initial guess */
float eps=1e5; /* error tolerance */
float f(float x);
float fd(float x);
printf("\nEnter the initial guess x0 ");
scanf("%f",&x0);
x1=x0;
do
{
k++;
x0=x1;
x1=x0f(x0)/fd(x0);
}while(fabs(x1x0)>eps);
printf("One root is %8.5f obtained at %d th iteration ",x1,k);
} /* main */
/* definition of the function f(x) */
float f(float x)
{
return(x*x*x3*x+1);
}
/* definition of the function fd(x) */
float fd(float x)
{
return(3*x*x3);
}
A sample of input/output:
Enter the initial guess x0 1.1
One root is 1.53209 obtained at 7 th iteration
4.10
Secant Method
f (xi )
f (xi )(xi xi1 )
= xi
.
f (xi )
f (xi ) f (xi1 )
(4.37)
This formula is same as the formula for Regulafalsi method and this formula needs
two initial guess x0 and x1 of the root.
Note 4.10.1 Regulafalsi method need an interval where the root belongs to, i.e., if
[x0 , x1 ] is the interval then f (x0 ) f (x1 ) < 0. But, secant method needs two nearest
values x0 and x1 of the exact root and not necessarily f (x0 ) f (x1 ) < 0.
Geometrical interpretation
A geometrical interpretation of secant method is illustrated in Figure 4.13.
A secant is drawn connecting f (xi1 ) and f (xi ). The point where it cuts the xaxis
is xi+1 . Another secant is drawn connecting f (xi ) and f (xi+1 ) to obtain xi+2 , and so
on.
4.10.1
(4.38)
Let be the exact root of the equation f (x) = 0 and the error at the nth iteration is
n = xn . Also f () = 0.
227
f (x)
6
xi
xi+1
xi+2
xi1
1
1
2n f ()
f ()
+ 1 + (n + n1 )
+
= n n +
2 f ()
2
f ()
f ()
1
+ O(2n n1 + n 2n1 ).
= n n1
2
f ()
n+1 = n
1 f ()
. This is a nonlinear dierence equation and
2 f ()
1/p
= Apn . Then n = Apn1 . This gives n1 = n A1/p .
.
Therefore, Apn = cn n A1/p , i.e., pn = cA(1+1/p) n
Equating the power of n on both sides, obtain the equation for p
1/p
p=1+
1+1/p
1
p
or
1
p = (1 5).
2
.
Positive sign gives p = 1.618. Hence n+1 = A1.618
n
Thus the rate of convergence of secant method is 1.618, which is smaller than the
NewtonRaphson method. Thus this method converges at a slower rate. However, this
method evaluates function only once in each iteration, but NewtonRaphson method
f (x0 )
1.0000
10.8750
0.1841
0.0333
x1
3.5000
3.0421
3.0497
3.0514
f (x1 )
10.8750
0.1841
0.0333
0.0005
x2
3.0421
3.0497
3.0514
3.0514
f (x2 )
0.1841
0.0333
0.0005
0.0005
229
Program 4.5
.
/* Program Secant
Program to find a root of the equation x*sin(x)1=0 by
secant method. It is assumed that a root lies between x0 and x1.*/
#include<stdio.h>
#include<math.h>
#include<stdlib.h>
void main()
{
float x0,x1,x2,fx0,fx1,fx2;
float eps=1e5; /* error tolerance */
float delta=1e5; /* slope */
float f(float x);
printf("\nEnter the values of x0 and x1 ");
scanf("%f %f",&x0,&x1);
fx0=f(x0);
fx1=f(x1);
if(fabs(fx1fx0)<delta){
printf("Slope too small
the method does not give correct root or fail");
exit(0);
}
do
{
x2=(x0*fx1x1*fx0)/(fx1fx0);
fx2=f(x2);
if(fabs(fx2)<eps){
printf("One root is %8.5f ",x2);
exit(0);
}
fx0=fx1; fx1=fx2;
x0=x1;
x1=x2;
}while(fabs(fx2)>eps);
} /* main */
/* definition of function f(x), it may change accordingly */
float f(float x)
{
return(x*sin(x)1);
}
4.11
Chebyshev Method
Let us consider the equation f (x) = 0. The function f (x) is expanded by Taylors series
in the neighbourhood of xn as 0 = f (x) = f (xn ) + (x xn )f (xn ) + .
f (xn )
.
This relation gives
x = xn
f (xn )
This is the (n + 1)th approximation to the root. Therefore,
xn+1 = xn
f (xn )
.
f (xn )
(4.39)
Again, expanding f (x) by Taylors series and retaining up to second order term,
shown below.
0 = f (x) = f (xn ) + (x xn )f (xn ) +
(x xn )2
f (xn )
2
(xn+1 xn )2
f (xn ) = 0.
2
Substituting the value of xn+1 xn from (4.39) to the last term and we nd
Therefore,
xn+1 = xn
1 [f (xn )]2
f (xn ) = 0.
2 [f (xn )]2
1 [f (xn )]2
f (xn )
f (xn ).
f (xn ) 2 [f (xn )]3
(4.40)
4.12
Muller Method
The main idea of this method is, the function f (x) is approximated by a quadratic
polynomial passing through the three points in the neighbour of the root. The root of
this quadratic is assumed to approximate the root of the equation f (x) = 0.
Let xn2 , xn1 , xn be any three distinct approximation to a root of the equation
f (x) = 0. We denote f (xn2 ) = fn2 , f (xn1 ) = fn1 and f (xn ) = fn .
231
(4.41)
Suppose, (4.41) passes through the points (xn2 , fn2 ), (xn1 , fn1 ) and (xn , fn ),
then
ax2n2 + bxn2 + c = fn2
(4.42)
ax2n1 + bxn1 + c =
ax2n + bxn + c = fn
(4.43)
fn1
(4.44)
f (x) =
(4.46)
since f (x) = 0.
Now, introducing
=
h
,
hn
n =
hn
hn1
and
n = 1 + n .
(4.47)
(4.48)
where
cn = n (n fn2 n fn1 + fn ).
The equation (4.48) now becomes
1
gn
+ cn = 0.
n fn 2 +
2 f
n n
.
gn gn2 4n fn cn
(4.49)
(4.50)
Now, replacing x on left hand side by xn+1 and obtain the formula
xn+1 = xn + (xn xn1 ),
(4.51)
hn = xn xn1 , n =
233
xn2
2.00000
2.50000
3.00000
2.28252
xn1
2.50000
3.00000
2.28252
2.27897
xn
3.00000
2.28252
2.27897
2.27902
gn
23.50000
3.89279
0.04477
0.00056
cn
3.75000
1.74262
0.00010
0.00000
1.43497
0.00494
0.01263
0.00338
xn+1
2.28252
2.27897
2.27902
2.27902
Formula
1. Bisection
xn+1 =
Order of Evaluation
Convergent of function
in each step
xn + xn1
2
xn1 fn xn fn1
fn fn1
2. False Position
xn+1 =
3. Iteration
xn+1 = (xn )
4. NewtonRaphson xn+1
5. Secant
Gain of one
f (xn )
= xn
f (xn )
xn+1 =
xn1 fn xn fn1
fn fn1
bit per
iteration
1
1.62
1.84
6. Modied
NewtonRaphson
xn+1 = xn
7. Chebyshev
xn+1 = xn
8. Muller
xn+1
f (xn
fn
12 fn /fn )
fn 1 fn2
f
fn
2 f 3n n
= xn + (xn xn1 )
1
,
f (x)
f (x)
.
2{f (x)}3
{f (x)}2
f (x)
+
.
6{f (x)}4 2{f (x)}5
f (x)
f (x)
f (x)f (x)
+
6
f (x)
2{f (x)}3
235
k is integer.
Find the value of k such that the iteration scheme has fast convergence.
Solution. Let lim xn = lim xn+1 = .
n
n
Since is a root of the given equation,
3 5 2 + 4 3 = 0.
Substituting, xn = + n , xn+1 = + n+1 to the given iteration scheme. Then
k(n+1 + ) = 3 + (k 4)(n + ) + 5(n + )2 (n + )3
kn+1 = (3 4 + 5 2 3 ) + n {(k 4) + 10 3 2 }
+2n (5 3) 3n
or, kn+1 = n {(k 4) + 10 3 2 } + 2n (5 3) 3n .
If k 4 + 10 3 2 = 0 or, k = 3 2 10 + 4 then
n+1 = A2n , where A =
1
(5 3)
k
af (xn )
f (xn bf (xn )/f (xn ))
where a, b are arbitrary parameters, for solving the equation f (x) = 0. Determine
a and b such that the iteration method is of order as high as possible for nding a
simple root of f (x) = 0.
af (x)
af (x)
=x
f (x bf (x)/f (x))
f (g(x))
where
g(x) = x
bf (x)
,
f (x)
g (x) = 1
(x) = 1
{f (g(x))}4
f ()f () f ()f ()(1 b)
() =
{f ()}2
f ()f ()(1 b)
f ()
{1 + 2(1 b)}.
+
=
{f ()}2
f ()
(x) = 1
2n
3
() + n () +
2
6
2n
3
() + n () +
2
6
237
3n
().
6
Hence the iteration scheme will have a third order convergence when a = 1 and
b = 1/2.
4.13
(4.52)
Domains of roots
(4.53)
where a0 , a1 , . . . , an are real coecients, and let A = max{a1 , a2 , . . . , an } and B =
max{a0 , a1 , . . . , an1 }. Then the roots of the equation (4.53) lie in the interval
r < x < R where
r=
1
1 + B/an 
and
R=1+
A
.
a0 
(4.54)
Here r is the lower bound and R is the upper bound of the positive roots of the
equation (4.53) and R and r are the lower and the upper bounds of the negative
roots respectively.
The Lagranges or Newtons method may also be used to nd the upper bound of the
positive roots of the equation (4.53).
Theorem 4.10 (Lagranges). If the coecients of the polynomial
a0 xn + a1 xn1 + + an1 x + an = 0
satisfy the conditions a0 > 0, a1 , a2 , . . . , am1 0, am <
0, for some m n, then the
upper bound of the positive roots of the equation is 1 + m B/a0 , where B is the greatest
of the absolute values of the negative coecients of the polynomial.
Theorem 4.11 (Newtons). If for x = c the polynomial
f (x) a0 xn + a1 xn1 + + an1 x + an = 0
and its derivatives f (x), f (x), . . . assume positive values then c is the upper bound of
the positive roots of the equation.
The roots of a polynomial equation can be determined in two techniques iteration
methods and direct methods. In this section, two iteration methods, viz., BirgeVieta
and Bairstow methods, and one direct method Graees root squaring method are
discussed.
239
Iterative Methods
4.14
BirgeVieta Method
This method is based on the NewtonRaphson method. Here a real number is determined such that (x ) is a factor of the polynomial
Pn (x) = xn + a1 xn1 + a2 xn2 + + an1 x + an = 0.
(4.55)
Let Qn1 (x) and R be the quotient and remainder when Pn (x) is divided by the
factor (x ), where Qn1 (x) is a polynomial of degree (n 1) of the form
Qn1 (x) = xn1 + b1 xn2 + b2 xn3 + + bn2 x + bn1 .
(4.56)
(4.57)
Thus
(4.58)
Pn (xk )
, k = 0, 1, 2, . . . .
Pn (xk )
(4.59)
The values of Pn (xk ) and Pn (xk ) can be determined by synthetic division. To
determine the values of b1 , b2 , . . . , bn1 and R, comparing the coecient of like powers
of x on both sides of (4.57) and obtain the following relations.
a1 = b1
a2 = b2 b1
..
.
b1 = a1 +
b2 = a2 + b1
..
.
ak = bk bk1
..
.
bk = ak + bk1
..
.
an = R bn1
R = an + bn1
From (4.57),
Pn () = R = bn (say).
(4.60)
(4.61)
(4.62)
+ + bn2 xi + bn1 .
Pn (xi ) = xin1 + b1 xn2
i
(4.63)
Thus
Pn (x) can be evaluated as Pn (x) is evaluated. Dierentiating (4.61) with respect to
and obtain
dbk1
dbk
= bk1 +
.
d
d
Let
dbk
= ck1 .
d
(4.64)
(4.65)
dbn
dR
=
= cn1
d
d
[using (4.64)].
bn
cn1
, k = 0, 1, 2, . . . .
(4.66)
241
a2
x0 b1
b2
x0 c1
c2
an2
x0 bn3
bn2
x0 cn3
cn2
an1
an
x0 bn2
x0 bn1
bn1
bn = R
x0 cn2
cn1 = Pn (x0 )
Example 4.14.1 Find all the roots of the polynomial equation x4 8x3 +14.91x2 +
9.54x 25.92 = 0. One root of the equation lies between 1 and 2.
Solution. Let the polynomial be denoted by P4 (x). Also, let the initial guess be
x0 = 1.2.
1.2 1
8 14.91 9.54
25.92
1.2 8.16 8.10
21.168
1.2 1 6.8 6.75 17.64
4.752=b4 = P4 (x0 )
1.2 6.72 0.036
1 5.6 0.03 17.676=c3 = P4 (x0 )
Therefore,
x1 = x0
b4
4.752
= 1.46884.
= 1.2
c3
17.676
1.46884 1
8
14.91 9.54
25.92
1.46884 9.59323 7.80949
25.48362
1.46884 1 6.53116 5.31677 17.34949
0.43638=b4
1.46884 7.43574 3.11243
1 5.06232 2.11897 14.23706=c3
Then x2 = x1
b4
0.43638
= 1.49949.
= 1.46884
c3
14.23706
1.49949 1
8
14.91 9.54
25.92
1.49949 9.74745 7.74119
25.91298
1.49949 1 6.50051 5.16255 17.28119
0.00702=b4
1.49949 7.49898 3.50345
1 5.00102 2.33643 13.77774=c3
Then x3 = x2
b4
0.00702
= 1.50000.
= 1.49949
c3
13.77774
5.16255
17.28119
10.00204 19.35796
4.83949
2.07677=b3
5.99796
1.15847= c2
b3
2.07677
= 5.79268.
=4
c2
1.15847
5.79268 1 6.50051
5.79268
5.79268 1 0.70783
5.79268
1 5.08485
x2 = x1
b3
23.43485
= 5.02476.
= 5.79268
c2
30.51723
5.02476 1 6.50051
5.02476
5.02476 1 1.47575
5.02476
1 3.54901
x3 = x2
5.16255
17.28119
7.41529
11.31948
2.25274
5.96171=b3
17.83292
15.58018= c2
b3
5.96171
= 4.64211.
= 5.02476
c2
15.58018
4.64211 1 6.50051
4.64211
4.64211 1 1.85840
4.64211
1 2.78371
x4 = x3
5.16255
17.28119
4.10023
6.15366
1.06232
23.43485=b3
29.45491
30.51723= c2
5.16255
17.28119
8.62690
16.08188
3.46435
1.19931=b3
12.92229
9.45794= c2
b3
1.19931
= 4.51531.
= 4.64211
c2
9.45794
243
Program 4.6
.
/* Program BirgeVieta for polynomial equation
Program to find a root of the polynomial equation by
BirgeVieta method. Leading coefficient is 1 */
#include<stdio.h>
#include<math.h>
#include<stdlib.h>
void main()
{
int n, N,i,k;
float x0,x1,a[10],b[10],c[10];
float epp=1e5; /* error tolerance */
printf("\nEnter the degree of the polynomial ");
scanf("%d",&n);
printf("Enter the coefficients of the polynomial,
except leading coeff.");
for(i=1;i<=n;i++) scanf("%f",&a[i]);
printf("Enter the initial guess x0 ");
scanf("%f",&x0);
printf("Enter maximum number of iterations to be done ");
scanf("%d",&N);
b[0]=1; c[0]=1;
for(i=1;i<=N;i++)
{
for(k=1;k<=n;k++) b[k]=a[k]+x0*b[k1];
for(k=1;k<=n1;k++) c[k]=b[k]+x0*c[k1];
x1=x0b[n]/c[n1];
if(fabs(x1x0)<epp)
{
printf("One root is %8.5f obtained at %d iterations",x1,i);
printf("\nCoefficients of the reduced polynomial are\n ");
for(k=0;k<=n1;k++) printf("%f ",b[k]);
exit(0);
}
else
x0=x1;
} /* i loop */
printf("\n Root not found at %d iterations ",N);
} /* main */
245
A sample of input/output:
Enter the degree of the polynomial 4
Enter the coefficients of the polynomial, except leading coeff.
3 0 6 4
Enter the initial guess x0 1
Enter maximum number of iterations to be done 100
One root is 1.00000 obtained at 1 iterations
Coefficients of the reduced polynomial are
1.000000 2.000000 2.000000 4.000000
4.15
Bairstow Method
(4.67)
(4.69)
and
S(pt , qt ) = S(p + p, q + q) = 0.
The derivatives are evaluated at (p, q). Neglecting the square and higher powers of
p and q the above equations reduce to
pRp + qRq = R
(4.71)
pSp + qSq = S.
(4.72)
SRp RSp
RSq SRq
, q =
.
Rp Sq Rq Sp
Rp Sq Rq Sp
(4.73)
Now, we determine the coecient of the polynomial Qn2 (x) and the expression for
R and S in terms of p and q.
From equations (4.67)(4.69)
xn + a1 xn1 + a2 xn2 + + an1 x + an
= (x2 + px + q)(xn2 + b1 xn3 + + bn3 x + bn2 ).
(4.74)
b1 = a1 p
b2 = a2 pb1 q
..
.
ak = bk + pbk1 + qbk2
..
.
bk = ak pbk1 qbk2
..
.
(4.75)
In general,
bk = ak pbk1 qbk2 ,
k = 1, 2, . . . , n
(4.76)
where b0 = 1, b1 = 0.
In this notation,
R = bn1 ,
S = bn + pbn1 .
(4.77)
247
Thus R and S are available when bs are known. To determine the partial derivatives
Rp , Rq , Sp and Sq , dierentiating (4.76) with respect to p and q.
bk
bk1
bk2
= bk1 p
q
,
p
p
p
bk2
bk1
bk
= bk2 p
q
,
q
q
q
Denoting
bk
= ck1 ,
p
bk
= ck2 .
and
q
b0
b1
=
=0
p
p
b0
b1
=
=0
q
q
k = 1, 2, . . . , n
(4.78)
(4.79)
(4.80)
(4.81)
(4.82)
(4.83)
(4.84)
bn1
= cn2
p
bn
bn1
Sp =
+p
+ bn1 = bn1 cn1 pcn2
p
p
bn1
= cn3
Rq =
q
bn1
bn
+p
= (cn2 + pcn3 ).
Sq =
q
q
Rp =
To nd the explicit expression for p and q, substituting the above values in (4.73).
Therefore,
bn cn3 bn1 cn2
cn3 (cn1 bn1 )
bn1 (cn1 bn1 ) bn cn2
.
q = 2
cn2 cn3 (cn1 bn1 )
p =
c2n2
(4.85)
and
q1 = q0 + q.
(4.86)
a1
p0
b1
p0
c1
p0
q0
p0
q0
a2
p0 b1
q0
b2
p0 c1
q0
c2
ak
p0 bk1
q0 bk2
bk
p0 ck1
q0 ck2
ck
an1
p0 bn2
q0 bn3
bn1
p0 cn2
q0 cn3
cn1
an
p0 bn1
q0 bn2
bn
Once p1 and q1 are evaluated, the next improved values p2 , q2 are determined from
the relation
p2 = p1 + p,
q2 = q1 + q.
In general,
pk+1 = pk + p,
qk+1 = qk + q,
(4.87)
4.00000
0.50000
1.00000
3.50000 = b1
0.50000
1.00000
3.00000
= c1
0.5
0.5
0.5
0.5
7.00000
1.75000
0.50000
9.25000
1.50000
0.50000
11.25000
= c2
22.00000
4.62500
1.75000
19.12500
5.62500
1.50000
15.00000
= c3
24.00000
9.56250
4.62500
38.18750 = b4
p =
c22
b4 c1 b3 c2
= 0.88095,
c1 (c3 b3 )
q =
b3 (c3 b3 ) b4 c2
= 3.07143
c22 c1 (c3 b3 )
4.00000
1.38095
1.00000
2.61905
1.38095
1.00000
1.23810
1.38095
2.57143
1.38095
2.57143
7.00000
3.61678
2.57143
8.04535
1.70975
2.57143
7.18367
22.00000
11.11025
6.73469
4.15506
9.92031
3.18367
8.94893
24.00000
5.73794
20.68805
9.04989
p = 0.52695, q = 0.29857.
p2 = p1 + p = 1.90790, q2 = q1 + q = 2.86999.
Third iteration
1.00000
1.90790
2.86999
1.00000
1.90790
2.86999
1.00000
4.00000
1.90790
2.86999
2.09210
1.90790
0.18420
7.00000
3.99152
6.00432
8.12152
0.35144
2.86999
5.60297
22.00000
15.49504
23.30873
0.50064
10.68990
0.52866
10.71793
24.00000
0.95517
1.64644
p = 0.08531, q = 0.12304.
p3 = p2 + p = 1.99321, q3 = q2 + q = 2.99304.
Fourth iteration
1.00000
4.00000
1.99321
1.00000
2.00679
1.99321
1.00000
0.01359
1.99321
2.99304
1.99321
2.99304
7.00000
3.99995
2.99304
8.00692
0.02709
2.99304
5.04097
p = 0.00676, q = 0.00692.
p4 = p3 + p = 1.99996, q4 = q3 + q = 2.99996.
22.00000
15.95943
6.00642
0.03416
10.04768
0.04067
10.05419
24.00000
0.06808
23.96501
0.10307
249
4.00000
1.99996
1.00000
2.00004
1.99996
1.00000
0.00008
1.99996
2.99996
1.99996
2.99996
7.00000
4.00000
2.99996
8.00004
0.00015
2.99996
5.00023
22.00000
15.99978
6.00004
0.00018
10.00027
0.00023
10.00031
24.00000
0.00037
23.99981
0.00055
p = 0.00004, q = 0.00004.
p5 = p4 + p = 2.00000, q5 = q4 + q = 3.00000.
Therefore, a quadratic factor is x2 + 2x 3 which is equal to (x 1)(x + 3). The
deated polynomial is Q2 (x) = x2 + 2.00004x 8.00004 x2 + 2x 8.
Thus P4 (x) = (x 1)(x + 3)(x2 + 2x 8) = (x 1)(x + 3)(x 2)(x + 4).
Hence the roots of the given equation are 1, 3, 2, 4.
Algorithm 4.7 (Bairstow method). This algorithm extracts a quadratic factor from a polynomial of degree n and also determines the deated polynomial, by
Bairstow method.
Algorithm Bairstow
// Extract a quadratic factor x2 + px + q from a polynomial Pn (x) = xn + a1 xn1 +
+ an1 x + an of degree n and determines the deated polynomial Qn2 (x) =
xn2 + b1 xn3 + b2 xn4 + + bn2 .//
//the degree and the coecients of the polynomial.//
Read n, a1 , a2 , . . . , an ;
Read p, q, ;
//the initial guess of p, q and error tolerance.//
Set b0 = 1, b1 = 0, c0 = 1, c1 = 0;
//Compute bk and ck
1. for k = 1 to n do
Compute bk = ak pbk1 qbk2 ;
endfor;
for k = 1 to n 1 do
Compute ck = bk pck1 qck2 ;
endfor;
(cn1 bn1 )bn cn2
cn3 bn1 cn2
;
q = bn1
Compute p = c2 bnc
c2n2 cn3 (cn1 bn1 ) ;
n3 (cn1 bn1 )
n2
Compute pnew = p + p, qnew = q + q;
if (pnew p < ) and (qnew q < ) then
Print The values of p and q are, pnew , qnew ;
Stop;
endif;
Set p = pnew , q = qnew ;
251
pnew=p+delp;
qnew=q+delq;
}while((fabs(pnewp)>epp  fabs(qnewq)>epp));
findroots(p,q);
n=2;
for(i=1;i<=n;i++) a[i]=b[i];
}while(n>2);
/* deflated polynomial is quadratic */
if(n==2) findroots(b[1],b[2]);
/* deflated polynomial is linear */
if(n==1) printf("%f ",b[1]);
} /* main */
/* finds the roots of the quadratic x*x+px+q=0 */
void findroots(float p, float q)
{
float dis;
dis=p*p4*q;
if(dis>=0)
{
printf("%f %f\n",(p+sqrt(dis))/2,(psqrt(dis))/2);
}
else
{
printf("(%f,%f), (%f,%f)\n",p/2,sqrt(fabs(dis))/2,
p/2,sqrt(fabs(dis))/2);
}
} /* findroots */
A sample of input/output:
Enter the degree of the polynomial 5
Enter the coefficients of the polynomial, except leading coeff.
3 4 5 6 1
The roots are
(0.604068,0.729697), (0.604068,0.729697)
(2.030660,1.861934), (2.030660,1.861934)
0.146816
253
Direct Method
4.16
This method may be used to nd all the roots of all types (real, equal or complex) of a
polynomial equation with real coecients. In this method, an equation is constructed
whose roots are squares of the roots of the given equation, then another equation whose
roots are squares of the roots of this new equation and so on, the process of rootsquaring
being continued as many times as necessary.
Let the given equation be
xn + a1 xn1 + a2 xn2 + + an1 x + an = 0
(4.88)
(4.89)
b1 = a21 2a2
b2 = a22 2a1 a3 + 2a4
..
.
bk = a2k 2ak1 ak+1 + 2ak2 ak+2
..
.
(4.90)
bn = a2n .
The roots of the equation (4.89) are 12 , 22 , . . . , n2 . The coecients bk s can be
obtained from Table 4.3.
The (k + 1)th column i.e., bk of Table 4.3 can be obtained as follows:
The terms alternate in sign starting with a positive sign. The rst term is a2k . The
second term is twice the product of ak1 and ak+1 . The third term is twice the product
a1
a21
2a2
a2
a22
2a1 a3
2a4
a3
a23
2a2 a4
2a1 a5
2a6
b1
b2
b3
a4
a24
2a3 a5
2a2 a6
a1 a7
2a8
b4
an
a2n
bn
of ak2 and ak+2 . This process is continued until there are no available coecients to
form the cross product terms.
The rootsquaring process is repeated to a sucient number of times, say m times
and we obtain the equation
xn + c1 xn1 + c2 xn2 + + cn1 x + cn = 0.
(4.91)
Let the roots of (4.91) be 1 , 2 , . . . , n . This roots are the 2m th power of the roots
of the equation (4.88) with opposite signs, i.e.,
m
i = i2 , i = 1, 2, . . . , n.
The relation between roots and coecients gives
p
c1 = 1 + 2 + + n =
i =
i
p p
c2 = 1 2 + 1 3 + + n1 n =
i j =
i j
c3 =
i<j<k
i j k =
i<j
i<j
ip jp kp
i<j<k
(1)n cn = 1 2 n = (1)n 1p 2p np , where p = 2m .
In the following dierent cases are considered separately.
Case I. Roots are real and unequal in magnitudes.
Let us consider
n  < n1  < < 2  < 1 
then
n  n1  2  1 .
(4.92)
255
That is,
m
n 2 n1 2 2 2 1 2
(4.93)
since all the roots are widely separated in magnitude at the nal stage.
Then from (4.92),
p p
p
2
3
n
p
c1 = 1 1 +
1p
+
+ + +
1
1
1
as
p
i
p
1 3
n1 n p
p
(1 2 )p
+ +
c2 = (1 2 ) 1 +
1 2
1 2
(4.94)
This determines the absolute values of the roots. By substituting these values in the
original equation (4.88) one can determine the sign of the roots. The squaring process is
terminated when another squaring process produces new coecients that are almost the
squares of the corresponding coecients ck s i.e., when the cross product terms become
negligible with respect to square terms. Thus the nal stage is identied by the fact
that on rootsquaring at that stage all the cross products will vanish.
Case II. All roots are real with one pair of equal magnitude.
Let 1 , 2 , . . . , n be the roots of the given equation, if a pair of roots are equal in
magnitude then this pair is conveniently called a double root. A double root can be
identied in the following way:
If the magnitude of the coecient ck is about half the square of the magnitude of the
corresponding coecient in the previous equation, then it indicates that k is a double
root. The double root is determined by the following process.
We have
ck
ck+1
k
and
k+1
.
ck1
ck
Then
c
k+1
k k+1 k2
.
ck1
m)
c
k+1
=
.
ck1
(4.95)
where p = 2 .
This gives the magnitude of the double root. The sign is determined by substituting
the root to the equation.
The double root can also be determined directly since k and k+1 converge to the
same root after sucient squaring. Generally, the rate of convergence to the double
root is slow.
Case III. One pair of complex roots and other roots are distinct in magnitude.
Let k and k+1 form a complex pair and let
k , k+1 = k eik
where k = k  = k+1 .
For suciently large m, k can be determined from the previous case,
c
ck+1
k+1
2p
2
k 
or,
, where p = 2m .
k
ck1
ck1
and k is determined from the relation
2m
k cos mk
ck+1
.
ck1
c
k+1
ck1
1/p
,
257
From this relation one can determine the value of u. Then the value of v can be
determined from the relation
v 2 = 2k u2 .
The presence of complex roots in (k + 1)th column is identied by the following
technique:
If the coecients of xnk in the successive squaring to uctuate both in magnitude and
sign, a complex pair can be detected by this oscillation.
Merits and Demerits
1. All roots are found at the end of the method, i.e., at one execution of the method
all the roots are determined, including complex roots.
2. No initial guess is required.
3. As a direct method, there is no scope for correcting the error generated in any
stage. If any error is generated at any stage, then the error propagates to all the
subsequent computations and ultimately gives a wrong result.
4. The method is laborious, and to get a very accurate result the method has to be
repeated for a large number of times.
5. There is a chance for data overow in computer.
In the following, three examples are considered to discuss the three possible cases of
Graees method. The following table is the Graees root squaring scheme for four
degree equation.
1
1
a1
a2
a3
a4
a21
a22
a23
a24
2a2 2a1 a3 2a2 a4
2a4
1 c1
c2
c3
c4
Example 4.16.1 Find the roots of the equation
2x4 15x3 + 40x2 45x + 18 = 0
correct up to four decimal places by Graees root squaring method.
Solution. All the calculations are shown in the following table. The number within
the parenthesis represents the exponent of the adjacent number. i.e., 0.75(02) means
0.75 102 .
m 2m
x4
x3
x2
0 1 1.00000 0.75000(01) 0.20000(02)
1.00000 0.56250(02) 0.40000(03)
0.40000(02) 0.33750(03)
0.18000(02)
1 2 1.00000 0.16250(02) 0.80500(02)
1.00000 0.26406(03) 0.64803(04)
0.16100(03) 0.47531(04)
0.16200(03)
2 4 1.00000 0.10306(03) 0.18891(04)
1.00000 0.10622(05) 0.35688(07)
0.37783(04) 0.17207(07)
0.13122(05)
3 8 1.00000 0.68436(04) 0.18612(07)
1.00000 0.46835(08) 0.34640(13)
0.37223(07) 0.61457(12)
0.86093(08)
4 16 1.00000 0.43113(08) 0.28495(13)
1.00000 0.18587(16) 0.81195(25)
0.56989(13) 0.16002(24)
0.37060(16)
5 32 1.0000 0.18530(16) 0.79595(25)
x
1
0.22500(02) 0.90000(01)
0.50625(03) 0.81000(02)
0.36000(03)
0.14625(03) 0.81000(02)
0.21389(05) 0.65610(04)
0.13041(05)
0.83481(04) 0.65610(04)
0.69690(08) 0.43047(08)
0.24789(08)
0.44901(08) 0.43047(08)
0.20161(16) 0.18530(16)
0.16023(15)
0.18559(16) 0.18530(16)
0.34443(31) 0.34337(31)
0.10560(29)
0.34337(31) 0.34337(31)
This is the nal equation since all the cross products vanish at the next step and all
the roots are real and distinct in magnitude.
Therefore,
1  = (0.18530 1016 )1/32 = 3.0000,
0.79595 1025 1/32
2  =
= 2.0000,
0.18530 1016
3  =
4  =
0.34337 1031
0.79595 1025
0.34337 1031
0.34337 1031
1/32
= 1.5000,
1/32
= 1.0000.
259
x
1
0.60000(01) 0.40000(01)
0.36000(02) 0.16000(02)
0.00000(00)
0.36000(02) 0.16000(02)
0.12960(04) 0.25600(03)
0.89600(03)
0.40000(03) 0.25600(03)
0.16000(06) 0.65536(05)
0.86016(05)
0.73984(05) 0.65536(05)
0.54736(10) 0.42950(10)
0.11450(10)
0.43286(10) 0.42950(10)
0.18737(20) 0.18447(20)
0.28936(18)
0.18447(20) 0.18447(20)
The diminishing double products vanish at the next step and hence this is the nal
equation and since we nd the characteristic behaviour of a double root in the second
column.
Therefore,
1  = (0.42951 1010 )1/32 = 2.0000,
0.18447 1020 1/64
2  = 3  =
= 1.4142,
0.42951 1010
0.18447 1020 1/32
4  =
= 1.0000.
0.18447 1020
Here 1.4142 as well as 1.4142 satised the given equation, hence the roots of the
given equation are 2, 1.4142, 1.
x
1
0.70000(01) 0.10000(02)
0.49000(02) 0.10000(03)
0.20000(02)
0.29000(02) 0.10000(03)
0.84100(03) 0.10000(05)
0.98000(04)
0.10641(05) 0.10000(05)
0.11323(09) 0.10000(09)
0.25340(08)
0.87891(08) 0.10000(09)
0.77248(16) 0.10000(17)
0.23437(16)
0.10069(17) 0.10000(17)
0.10137(33) 0.10000(33)
0.13733(31)
0.10000(33) 0.10000(33)
0.10000 1033
0.23283 1023
1/32
= 2.0000.
If 2
, 3 = u v, then 2u + 5 + 1 = 5 (sum of the roots). Therefore, u = 0.5. Then
v = 22 u2 = 2 0.25 = 1.3229.
Hence the roots are 5, 1, 0.5 1.3229i.
4.17
261
To solve a system of nonlinear equations the following methods are discussed in this
section.
1. The method of iteration (xed point iteration)
2. Seidal iteration
3. NewtonRaphson method.
4.17.1
g(x, y) = 0.
(4.96)
whose real roots are required within a specied accuracy. The above system can be
rewritten as
x = F (x, y)
and
y = G(x, y).
(4.97)
y1 = G(x0 , y0 )
x2 = F (x1 , y1 ),
y2 = G(x1 , y1 )
(4.98)
and
lim yn =
then
= F (, )
and
= G(, ).
(4.99)
Like the iteration process for single variable, the above sequence surely converge to
a root under certain condition. The sucient condition is stated below.
Seidal method
An improvement of the iteration method can be made by using the recently computed
values of xi while computing yi , i.e., xi+1 is used in the calculation of yi . Therefore, the
iteration scheme becomes
xn+1 = F (xn , yn )
yn+1 = G(xn+1 , yn ).
(4.101)
This method is called Seidal iteration. In case of three variables the scheme is
xn+1 = F (xn , yn , zn )
yn+1 = G(xn+1 , yn , zn )
(4.102)
2x x2 + 4y y 2 + 3
8x 4x2 + y 2 + 1
and y =
8
4
starting with (x0 , y0 ) = (1.1, 2.0), using (i) iteration method, and (ii) Seidal iteration
method.
263
Solution.
(i) Iteration method
Let
F (x, y) =
8x 4x2 + y 2 + 1
8
and
G(x, y) =
2x x2 + 4y y 2 + 3
.
4
xn+1 = F (xn , yn ) =
and yn+1
The value of xn , yn , xn+1 and yn+1 for n = 0, 1, . . . are shown in the following table.
n
0
1
2
3
4
xn
1.10000
1.12000
1.11655
1.11641
1.11653
yn
2.00000
1.99750
1.99640
1.99660
1.99661
xn+1
1.12000
1.11655
1.11641
1.11653
1.11652
yn+1
1.99750
1.99640
1.99660
1.99661
1.99660
xn+1 = F (xn , yn ) =
and
yn+1
xn
1.10000
1.12000
1.11600
1.11659
yn
2.00000
1.99640
1.99663
1.99660
xn+1
1.12000
1.11600
1.11659
1.11650
yn+1
1.99640
1.99663
1.99660
1.99660
Algorithm 4.8 (Seidal iteration). This algorithm used to solve two nonlinear
equations by Seidal iteration, when the initial guess is given.
Algorithm SeidalIteration2D
// Let (x0 , y0 ) be the initial guess of the system of equations x = F (x, y), y = G(x, y).
be the error tolerance, maxiteration represents the maximum number of repetitions
to be done.//
Input functions F (x, y), G(x, y).
Read , maxiteration, x0 , y0 , z0 ;
Set k = 0, error = 1;
While k < maxiteration and error > do
Set k = k + 1;
Compute x1 = F (x0 , y0 ), y1 = G(x1 , y0 );
Compute error = x1 x0  + y1 y0 ;
Set x0 = x1 , y0 = y1 ;
endwhile;
if error < then
Print The sequence converge to the root, x1 , y1 ;
Stop;
else
Print The iteration did not converge after, k,iterations;
Stop;
endif;
end SeidalIteration2D
Program 4.8
.
/* Program Seidal for a pair of nonlinear equations
Program to find a root of a pair of nonlinear equations
by Seidal method. Assumed that the equations are given
in the form x=f(x,y) and y=g(x,y).
The equations taken are x*x+4y*y4=0, x*x2xy+1=0. */
#include<stdio.h>
#include<math.h>
#include<stdlib.h>
void main()
{
int k=0,maxiteration;
float error=1,eps=1e5,x0,y0; /*initial guesses for x and y*/
float x1,y1;
float f(float x, float y);
265
NewtonRaphson method
(4.103)
(4.104)
(x0 ,y0 )
(4.105)
(x0 ,y0 )
Neglecting square and higher order terms, the above equations simplied as
f0
f0
+k
= f0
x
y
g0
g0
+k
= g0
h
x
y
f0 f
=
etc.
x
x (x0 ,y0 )
The above system can be written as
where f0 = f (x0 , y0 ),
f
0
x
g
0
x
f0
y h
f
0
1 f0
=
or
=
J
.
g0
g0
k
g k
0
(4.106)
f0
x
, where J =
g
0
x
f0
y
.
g0
y
267
(4.107)
Thus h and k are determined by one of the above two ways. Therefore, the new
approximations are then given by
x1 = x0 + h,
y1 = y0 + k.
(4.108)
The process is to be repeated until the roots are achieved to the desired accuracy.
The general formula is xn+1 = xn + h, yn+1 = yn + k; h, k are evaluated at (xn , yn )
instead at (x0 , y0 ).
If the iteration converges (the condition is stated below) then the rate of convergence
is quadratic.
Theorem 4.13 Let (x0 , y0 ) be an initial guess to a root (, ) of the system f (x, y) =
0, g(x, y) = 0 in a closed neighbourhood R containing (, ). If
1. f, g and their rst order partial derivatives are continuous and bounded in R, and
2. J = 0 in R, then the sequence of approximation xn+1 = xn + h, yn+1 = yn + k,
where h and k are given by (4.107), converges to the root (, ).
Example 4.17.2 Use NewtonRaphson method to solve the system x2 2x y +
0.5 = 0, x2 + 4y 2 4 = 0 with the starting value (x0 , y0 ) = (2.00, 0.25).
Solution. Let f (x, y) = x2 2x y + 0.5 and g(x, y) = x2 + 4y 2 4.
f
g
g
f
= 2x 2,
= 1,
= 2x,
= 8y.
x
y
x
y
Therefore,
(
J=
At (x0 , y0 ), J0 =
Therefore,
f
x
g
x
f
y
g
y
2x 2 1
,
2x
8y
f0
g0
2 1
.
4 2
2 1
4 2
h
k
0.25
0.25
0.25
.
0.25
h
k
1
=
8
2 1
4 2
0.25
0.25
0.09375
0.06250
1.81250 1.00000
f1
0.00879
,
At (x1 , y1 ), J1 =
=
.
g1
3.81250 2.50000
0.02441
h
f1
=
J1
k
g1
1
2.50000 1.00000
h
0.00879
0.00556
or,
=
=
.
k
0.02441
0.00129
8.34375 3.81250 1.81250
Therefore, x2 = x1 + h = 1.90625 0.00556 = 1.90069,
0.00129 = 0.31121.
y2 = y1 + k = 0.31250
0.00003
1.80138 1.00000
f2
=
.
,
At (x2 , y2 ), J2 =
0.00003
g2
3.80138 2.48968
h
f2
=
J2
k
g2
1
h
2.48968 1.00000
0.00003
0.00001
or,
=
=
.
k
0.00003
0.00001
8.28624 3.80138 1.80138
Hence, x3 = x2 + h = 1.90069 0.00001 = 1.90068,
y3 = y2 + k = 0.31121 + 0.00001 = 0.31122.
Thus, one root is x = 1.9007, y = 0.3112 correct up to four decimal places.
Algorithm 4.9 (NewtonRaphson method for pair of equations). This algorithm solves a pair of nonlinear equations by NewtonRaphson method. The initial
guess of a root is to be supplied.
Algorithm NewtonRaphson 2D
//(x0 , y0 ) is initial guess, is the error tolerance.//
Input functions f (x, y), g(x, y), fx (x, y), fy (x, y), gx (x, y), gy (x, y).
Read x0 , y0 , , maxiteration;
for i = 1 to maxiteration do
Compute f0 = f (x0 , y0 ), g0 = g(x0 , y0 );
if (f0  < and g0  < ) then
Print A root is, x0 , y0 ;
Stop;
endif;
Compute delf x =
f
x
,
(x0 ,y0 )
delf y =
f
y
(x0 ,y0 )
Compute delgx =
g
x
, delgy =
(x0 ,y0 )
g
y
(x0 ,y0 )
269
271
4.18
Exercise
1
+ sin xn , x0 = 1
2
3
(i) 1/15, (ii) 11, (iii) 5.
15. Show that an iterative method for computing
xn+1
and also deduce that
a is given by
1
a
(k 1)xn + k1
=
k
xn
k1 2
n+1
2ka n
where n is the error at the nth iteration. What is the order of convergence for
this iterative method ?
273
3 loge xn exn
p
x0 f (xn ) xn f (x0 )
f (xn ) f (x0 )
f (xn )
f (xn + f (xn )) f (xn )
, g(xn ) =
, n = 0, 1, 2, . . . .
g(xn )
f (xn )
26. Show that the iteration scheme to nd the value of a using Chebyshev third
order method is given by
1
a
a 2
1
.
xn+1 =
xn
xn +
2
xn
8xn
xn
xn+1 =
1 4 1 3
x x + 8xn 12.
16 n 2 n
Chapter 5
(5.1)
where
A=
a11
a21
ai1
am1
a12
a22
ai2
am2
AX = b,
b1
x1
a1n
b2
x2
a2n
..
..
. and X = . .
,
b
=
bi
xi
ain
..
.
.
.
amn
bm
xm
275
(5.3)
(5.4)
Direct Methods
5.1
Cramers Rule
To solve a system of linear equations, a simple method (but, not ecient) was discovered
by Gabriel Cramer in 1750.
Let the determinant of the coecients of the system (5.2) be D = aij ; i, j =
1, 2, . . . , n, i.e., D = A. In this method, it is assumed that D = 0. The Cramers
rule is described in the following. From the properties of determinant
a11 a12 a1n x1 a11 a12 a1n
a21 a22 a2n x1 a21 a22 a2n
x1 D = x1
=
an1 an2 ann x1 an1 an2 ann
a11 x1 + a12 x2 + + a1n xn a12 a1n
a21 x1 + a22 x2 + + a2n xn a22 a2n [Using the operation
=
C1 = C1 + x2 C2 + + xn Cn .]
an1 x1 + an2 x2 + + ann xn an2 ann
277
a1n
a2n
[Using (5.1)]
ann
b1 a1 i+1
b2 a2 i+1
bn an i+1
a1n
a2n
,
ann
i = 1, 2, , n
Example 5.1.1 Use Cramers rule to solve the following systems of equations
x1 + x2 + x3 = 2
2x1 + x2 x3 = 5
x1 + 3x2 + 2x3 = 5.
Solution. The determinant D of the system is
1 1 1
D = 2 1 1 = 5.
1 3 2
The determinants D1 , D2 and D3 are shown below:
2 1 1
1 2 1
D1 = 5 1 1 = 5,
D2 = 2 5 1 = 10,
5 3 2
1 5 2
1 1 2
D3 = 2 1 5 = 5.
1 3 5
5
10
5
D1
D2
D3
= = 1, x2 =
=
= 2, x3 =
= = 1.
D
5
D
5
D
5
Therefore the solution is x1 = 1, x2 = 2, x3 = 1.
Thus, x1 =
5.1.1
It may be noted that the Cramers rule involves to compute (n + 1) determinants each
of order n (for a system of n equations and n variables). Again, the Laplaces expansion
5.2
Evaluation of Determinant
a12
(1)
a22
0
a13
(1)
a23
(2)
a33
a1n
(1)
a2n
(2)
a3n ,
(n1)
ann
where
(k1)
(k)
(k1)
aij = aij
aik
(k1)
a
;
(k1) kj
akk
(5.5)
(0)
(n1)
(k)
(k1)
is zero
of aij one division is required. If akk
(k1)
If akk is small then the division leads to the loss
then further
of signicant
reduction is not possible.
digits. To prevent the loss of signicant digits, the pivoting techniques are used.
A pivot is the largest magnitude element in a row or a column or the principal
diagonal or the leading or trailing submatrix of order i (2 i n).
5
10
A=
20
3
279
1 0 5
8 3 10
3 30 8
50 9 10
20 is the pivot for the rst column, 30 is the pivot for the principal diagonal,
50
is the
5 1
pivot for this matrix and 10 is the pivot for the trailing submatrix
.
10 8
In the elementary row operation, if the any one of the pivot element is zero or very
small relative to other elements in that row then we rearrange the remaining rows in
such a way that the pivot becomes nonzero or not a very small number. The method is
called pivoting. The pivoting are of two types partial pivoting and complete pivoting.
Partial and complete pivoting are discussed in the following.
Partial pivoting
In the rst stage, the rst pivot is determined by nding the largest element in magnitude among the elements of rst column and let it be ai1 . Then rows i and 1 are
interchanged. In the second stage, the second pivot is determined by nding the largest
element in magnitude among the elements of second column leaving rst element and
let it be aj2 . The second and jth rows are interchanged. This process is repeated for
(n 1)th times. In general, at ith stage, the smallest index j is chosen for which
(k)
(k)
(k)
(k)
(k)
(k)
or
n
for i = 1, 2, . . . , n.
(5.6)
j=1
j=i
1 7 3
A = 4 5 1.
8 1 6
The largest element (in magnitude) in the rst column is 8. Then interchanging
rst and third rows i.e.,
8 1 6
A 4 5 1.
1 7 3
The largest element in second column leaving the rst row is 7, so interchanging
second and third rows.
The matrix after partial pivoting is
8 1 6
A 1 7 3.
4 5 1
1 7 3
Let us consider the matrix B = 4 8 5 to illustrate the complete pivoting. The
2 6 1
largest element (in magnitude) is determined among all the elements of the matrix. It
is 8 attained at the (2, 2) position. Therefore, rst and second columns, and rst and
second rows are interchanged. The matrix transferred to
8 4 5
B 7 1 3.
6 2 1
The number 8 is the rst pivot.
To nd second pivot, largest element is determined
1 3
from the trailing submatrix
. The largest element is 3 and it is at the position
2 1
(2, 3).
281
Interchanging
second and third columns. The nal matrix (after complete pivoting)
8 5 4
is 7 3 1 .
6 1 2
Example 5.2.1 Compute the determinant of the following matrix by a triangularization algorithm using (i) partial pivoting, and (ii) complete pivoting:
2 0 4
A = 4 6 1.
5 1 2
Solution. (i) The largest element in the rst column is 5, which is the rst pivot of
A.
Interchanging rst and third rows, we obtain
5 1 2
4 6 1.
2 0 4
and sign = 1.
2
4
Adding times the rst row to the second row, times the rst row to the third
5
5
4
2
row i.e., R2 = R2 R1 and R3 = R3 R1 , we get
5
5
5 1
2
0 26/5 13/5 .
0 2/5 24/5
26
, which is the largest element (in magnitude) among the
5
elements of second column except rst row. Since this element is in the (2,2) position,
so no interchange of rows is required.
2/5
2/5
times the second row to the third row i.e., R3 = R3 +
R2 we obtain
Adding
26/5
26/5
5 1
2
0 26/5 13/5 .
0 0
5
The second pivot is
6 4 1
0 2 4.
1 5 2
Adding 16 times the rst row to the third row i.e., R3 = R3 16 R1 we obtain
6 4
1
0 2
4 .
0 13/3 13/6
2
4
The pivot of the trailing submatrix
is 13/3. Interchanging the second
13/3 13/6
6 1
4
and third rows, we have 0 13/3 13/6 and sign = 1.
0 2
4
6
2
times the second row to the third row i.e., R3 = R3 R2 we obtain
Adding
13/3
13
6 1
4
0 13/3 13/6 .
0 0
5
Therefore, A = sign (6)(13/3)(5) = 130.
The algorithm for triangularization i.e., to nd the value of a determinant using
partial and complete pivoting are presented below.
Algorithm 5.1 (Evaluation of determinant using partial pivoting). This
algorithm nds the value of a determinant of order n n using partial pivoting.
Algorithm Det Partial Pivoting.
//The value of determinant using partial pivoting.//
Let A = [aij ] be an n n matrix.
Step 1. Read the matrix A = [aij ], i, j = 1, 2, . . . , n.
Step 2. Set k = 1 and sign = 1// sign indicates the sign of the determinant
when interchanges two rows.//
Step 3. Find a pivot from the elements akk , ak+1k , , ank in the kth
That is, ajk  =
column of A, and let ajk be the pivot.
max{akk , ak+1 k , . . . , ank }.
Step 4. If ajk = 0 then A = 0; print the value of A and Stop.
Step 5. If j = k then go to Step 6.
Otherwise interchange the kth and jth rows and set sign = sign.
283
ajk
times the kth row from the jth row for j = k+1, k+2, . . . , n
akk
ajk
.Rk , where
i.e., for j = k + 1, k + 2, . . . , n do the following Rj = Rj
akk
Rj , Rj are the old and new jth rows respectively.
// This step makes ak+1k , ak+2k , , ank zero.//
Step 7. Increment k by 1 i.e., set k = k+1. If k < n then goto Step 3. Otherwise,
//Triangularization is complete.//
Compute A = sign product of diagonal elements.
Print A and Stop.
end Det Partial Pivoting.
Step 6. Subtract
Program 5.1
.
/* Program Partial Pivoting
Program to find the value of a determinant using partial pivoting */
#include<stdio.h>
#include<stdlib.h>
#include<math.h>
void main()
{
int n,k,i,j,sign=1;
float a[10][10],b[10],prod,temp;
int max1(float b[],int k, int n);
printf("\nEnter the size of the determinant ");
scanf("%d",&n);
printf("Enter the elements rowwise ");
for(i=1;i<=n;i++) for(j=1;j<=n;j++)
scanf("%f",&a[i][j]);
for(k=1;k<=n;k++)
{
for(i=k;i<=n;i++) b[i]=a[i][k];
/* copy from a[k][k] to a[n][k] into b */
j=max1(b,k,n); /* finds pivot position */
if(a[j][k]==0)
{
printf("The value of determinant is 0");
exit(0);
}
if(j!=k) /* interchange k and j rows */
{
sign=sign;
285
Step 3. Find
a pivot from the elements of the trailing submatrix
ank ank+1 ann
i.e., apq  = max{akk , . . . , akn ; ak+1k , . . . , ak+1n ; ank , . . . , ann }.
Step 4. If apq = 0 then A = 0; print the value of A and Stop.
Step 5. If p = k then goto Step 6. Otherwise, interchange the kth and the pth
rows and set sign = sign.
Step 6. If q = k then goto Step 7. Otherwise, interchange the kth and the qth
columns and set sign = sign.
ajk
times the kth row to the jth row for j = k + 1, k + 2, . . . , n
Step 7. Subtract
akk
i.e., for j = k + 1, k + 2, . . . , n do the following
ajk
.Rk , where Rj , Rj are the old and new jth rows respecRj = Rj
akk
tively.
Step 8. Increment k by 1 i.e., set k = k + 1.
If k < n then goto Step 3. Otherwise,
compute A = sign product of diagonal elements.
Print A and Stop.
end Det Complete Pivoting.
Program 5.2
.
/*Program Complete Pivoting
Program to find the value of a determinant using complete pivoting */
#include<stdio.h>
#include<stdlib.h>
#include<math.h>
void main()
{
int n,k,i,j,sign=1,p,q;
float a[10][10],prod,max,temp;
printf("\nEnter the size of the determinant ");
scanf("%d",&n);
printf("Enter the elements rowwise ");
for(i=1;i<=n;i++) for(j=1;j<=n;j++) scanf("%f",&a[i][j]);
for(k=1;k<=n;k++)
{
/* finds the position of the pivot element */
max=fabs(a[k][k]); p=k; q=k; /* set initial maximum */
for(i=k;i<=n;i++)
for(j=k;j<=n;j++)
if(max<fabs(a[i][j])) { max=fabs(a[i][j]); p=i; q=j;}
if(a[p][q]==0)
{
printf("The value of determinant is 0");
exit(0);
}
if(p!=k) /* interchange k and p rows */
{
sign=sign;
for(i=1;i<=n;i++)
{
temp=a[p][i]; a[p][i]=a[k][i]; a[k][i]=temp;
}
}
if(q!=k) /* interchange k and q columns */
{
sign=sign;
for(i=1;i<=n;i++)
{
temp=a[i][q]; a[i][q]=a[i][k]; a[i][k]=temp;
}
}
for(j=k+1;j<=n;j++) /* makes a[k+1][k] to a[n][k] zero */
{
temp=a[j][k]/a[k][k];
for(i=1;i<=n;i++)
a[j][i]=temp*a[k][i];
}
} /* end of k loop */
prod=sign;
for(i=1;i<=n;i++) /* product of diagonal elements*/
prod*=a[i][i];
printf("The value of the determinant is %f ",prod);
}/* main */
A sample of input/output:
Enter the size of the determinant 4
Enter the elements rowwise
287
2 3 8 4
6 1 0 5
8 3 1 2
3 8 7 10
The value of the determinant is 1273.000000
Advantages and disadvantages of partial and complete pivoting
The general disadvantages of the pivoting is that the symmetry or regularity of the
original matrix may be lost. Partial pivoting requires less time in terms of interchanges
and search for the pivot than the complete pivoting. A combination of partial and
complete pivoting is expected to be very eective not only for computing a determinant
but also for solving system of linear equations. The pivoting brings in stability where a
method becomes unstable for a problem. The pivoting reduces the error due to the loss
of signicant digits.
5.3
Inverse of a Matrix
From the theory of matrices, it is well known that every square nonsingular matrix has
unique inverse. The inverse of a matrix A is dened by
A1 =
adj A
.
A
A11
A12
adj A =
A1n
(5.7)
and dened as
A21 An1
A22 An2
,
A2n Ann
GaussJordan Method
In this method, the given matrix A is augmented with a unit matrix of same size, i.e.,
.
if the order of A is n n then the order of the augmented matrix [A..I] will be n 2n.
a a a1n
11 12
.
a a a2n
[A..I] = 21 22
..
.
..
.
..
.
..
.
0 0
1 0
.
(5.8)
0 1
Then the inverse of A is computed in two stages. In the rst stage, A is converted
into an upper triangular form, using only elementary row operations (Gauss elimination
method discussed in Section 5.5). In the second stage, the upper triangular matrix
(obtained in rst stage) is reduced to an identity matrix by row operations. All these
.
operations are operated on the augmented matrix [A..I]. After completion of these
.
.
stages, the augmented matrix [A..I] is turned to [I..A1 ], i.e., the inverse of A is obtained
from the right half of augmented matrix.
Thus
.
.. Gauss Jordan
I..A1 .
A.I
At the end of the operations the matrix shown in (5.8) reduces to the following form:
.
1 0 0 .. a11 a12 a1n
..
a
0
1
0
a
.
a
21 22
2n .
(5.9)
...
..
0 0 1 . a a a
n1
n2
nn
2 4 5
Example 5.3.1 Find the inverse of the following matrix A = 1 1 2 .
3 4 5
.
Solution. The augmented matrix [A..I] can be written as
.
4 5 .. 1 0 0
.
..
.
[A..I] =
1
1
2
.
0
1
0
..
3 4 5 . 0 0 1
2
(5.10)
289
.
4 5 .. 0 0 1
..
.
1
1
2
.
0
1
0
..
2 4 5 . 1 0 0
.
1 4/3 5/3 .. 0 0 1/3
.
1
1 1 2 .. 0 1 0 R1 = 3 R1
.
2 4 5 .. 1 0 0
.
1 4/3 5/3 .. 0 0 1/3
..
R = R2 R1 ; R = R3 2R1
0
7/3
1/3
.
0
1
1/3
3
2
..
0 4/3 5/3 . 1 0 2/3
3
7
(The largest element (in magnitude) in the second column is , which is at the a22
3
position and so there is no need to interchange any rows).
.
1 4/3 5/3 .. 0 0
1/3
.
3
4
.
1 4/3 5/3 .. 0
0
1/3
.
7
.
1 0 13/7 .. 0
4/7 1/7
.
4
.
1 0 0 .. 1
0
1
.
13
1
1
0
1
1/13 5/13 1/13 .
7/13 4/13 6/13
Algorithm 5.3 (Matrix inverse). The following algorithm computes the inverse
of a nonsingular square matrix of order n n and if the matrix is singular it prints
the message the matrix is singular and hence not invertible.
Algorithm Matrix Inversion (using partial pivoting).
Let A = [aij ] be an n n matrix.
Step 1. Read the matrix A = [aij ], i, j = 1, 2, . . . , n.
Step 2. //Augment the matrix A.//
Augment the matrix A by a unit matrix of order n n. The resultant
matrix A becomes of order n 2n.
0, for i = j
1, for i = j
for i, j = 1, 2, . . . , n.
i.e., ai n+j =
Program 5.3
.
/* Program Matrix Inverse
Program to find the inverse of a square matrix using
partial pivoting */
#include<stdio.h>
#include<stdlib.h>
#include<math.h>
#define zero 0.00001
void main()
{
int n,m,k,i,j;
float a[10][20],temp;
printf("\nEnter the size of the matrix ");
scanf("%d",&n);
printf("Enter the elements rowwise ");
for(i=1;i<=n;i++) for(j=1;j<=n;j++) scanf("%f",&a[i][j]);
/* augment the matrix A */
for(i=1;i<=n;i++) for(j=1;j<=n;j++) a[i][n+j]=0;
for(i=1;i<=n;i++) a[i][n+i]=1;
m=2*n;
for(k=1;k<=n;k++)
{
/* finds pivot element and its position */
temp=fabs(a[k][k]); j=k; /* initial maximum */
for(i=k+1;i<=n;i++)
if(temp<fabs(a[i][k])){
temp=fabs(a[i][k]); j=i;
}
if(fabs(a[j][k])<=zero) /* if a[j][k]=0 */
{
printf("The matrix is singular and is not invertible");
exit(0);
}
if(j!=k) /* interchange k and j rows */
{
for(i=1;i<=m;i++){
temp=a[j][i]; a[j][i]=a[k][i]; a[k][i]=temp;
}
}
291
if(a[k][k]!=1)
{
temp=a[k][k];
for(i=1;i<=m;i++) a[k][i]/=temp;
}
for(j=k+1;j<=n;j++) /* makes a[k+1][k] to a[n][k] zero */
{
temp=a[j][k];
for(i=1;i<=m;i++)
a[j][i]=temp*a[k][i];
}
} /* end of k loop */
/* make left half of A to a unit matrix */
for(k=2;k<=n;k++)
{
for(j=k1;j>=1;j)
{
temp=a[j][k];
for(i=1;i<=m;i++) a[j][i]=temp*a[k][i];
}
}
printf("\nThe inverse matrix is \n");
for(i=1;i<=n;i++)
{
for(j=n+1;j<=m;j++)
printf("%f ",a[i][j]); printf("\n");
}
}/* main */
A sample of input/output:
Enter the size of the matrix 3
Enter the elements rowwise
0 1 2
3 2 1
4 3 2
The inverse matrix is
0.218750 0.125000 0.156250
0.062500 0.250000 0.187500
0.531250 0.125000 0.093750
293
operations.
Similarly, Stage II takes O(n3 ) time. Hence the time complexity to compute the
inverse of a nonsingular matrix is O(n3 ).
Since the Stage I is similar to the algorithm Det Partial Pivoting, so the time
complexity to compute the determinant is O(n3 ).
5.4
The system of equations (5.1) can be written in the matrix form (5.3) as
Ax = b
where A, b and x are dened in (5.4).
The solution of Ax = b is given
x = A1 b,
(5.11)
1 2 3
x
10
A = 1 3 2 ,
x = y ,
b = 7 .
2 1 1
z
5
1 2 3
Now, A = 1 3 2 = 30 = 0.
2 1 1
That is, A is nonsingular and hence A1 exists.
1 5 13
5.
adj A = 5 5
7 5
1
1 5 13
1
adj A
5 5
5.
=
Thus, A1 =
A
30 7 5
1
1 5 13
90
10
3
1
1
1
5 5
5
60 = 2 .
7 =
Therefore, x = A b =
30 7 5
30 30
1
5
1
Hence the required solution is x = 3, y = 2, z = 1.
Algorithm 5.4 (Matrix inverse method). This algorithm is used to solve a
system of linear equations Ax = b, where A = [aij ]nn , x = (x1 , x2 , . . . , xn )t ,
b = (b1 , b2 , . . . , bn )t , by matrix inverse method.
Algorithm Matrix Inverse Method
Step 1. Read the coecient matrix A = [aij ], i, j = 1, 2, . . . , n and the right
hand vector b = (b1 , b2 , . . . , bn )t .
Step 2. Compute the inverse of A by the algorithm Matrix Inverse.
Step 3. If A is invertible then compute x = A1 b and print x =
(x1 , x2 , . . . , xn )t .
Otherwise, print A is singular and the system has either no solution or
has innitely many solutions.
end Matrix Inverse Method
Program 5.4
.
/* Program Matrix Inverse Method
Program to find the solution of a system of linear
equation by matrix inverse method. Partial pivoting
is used to find matrix inverse. */
#include<stdio.h>
#include<stdlib.h>
#include<math.h>
#define zero 0.00001
float a[10][20],ai[10][10];
int n;
int matinv();
void main()
{
int i,j;
float b[10],x[10];
295
if(fabs(a[j][k])<=zero) /* if a[j][k]=0 */
{
printf("The matrix is singular and is not invertible");
return(0);
}
if(j!=k) /* interchange k and j rows */
{
for(i=1;i<=m;i++)
{
temp=a[j][i]; a[j][i]=a[k][i]; a[k][i]=temp;
}
}
if(a[k][k]!=1)
{
temp=a[k][k];
for(i=1;i<=m;i++) a[k][i]/=temp;
}
for(j=k+1;j<=n;j++) /* makes a[k+1][k] to a[n][k] zero */
{
temp=a[j][k];
for(i=1;i<=m;i++)
a[j][i]=temp*a[k][i];
}
} /* end of k loop */
/* make left half of A to a unit matrix */
for(k=2;k<=n;k++)
for(j=k1;j>=1;j){
temp=a[j][k];
for(i=1;i<=m;i++) a[j][i]=temp*a[k][i];
}
printf("\nThe inverse matrix is \n");
for(i=1;i<=n;i++)
for(j=n+1;j<=m;j++) ai[i][jn]=a[i][j];
return(1);
}/* matinv */
A sample of input/output:
Enter the size of the matrix 3
Enter the elements rowwise
297
0 1 2
3 2 1
4 3 2
The inverse matrix is
0.218750 0.125000 0.156250
0.062500 0.250000 0.187500
0.531250 0.125000 0.093750
Complexity
The time complexity of this method is O(n3 ) as the method involves computation of
A1 .
5.5
(1)
(1)
(1)
(1)
(1)
(1)
(1)
(5.12)
(1)
an2 x2
where
(1)
(1)
(1)
+ an3 x3 + + a(1)
nn xn = bn ,
aij = aij
ai1
a1j ; i, j = 2, 3, . . . , n.
a11
Again, to eliminate x2 from the third, forth, . . ., and nth equations the second equa(1)
(1)
(1)
a
a42
an2
(1)
tion is multiplied by 32
,
,
.
.
.,
(1)
(1)
(1)
(2)
(2)
(2)
(5.13)
(2)
(2)
an3 x3 + + a(2)
nn xn = bn ,
where
(1)
(2)
(1)
aij = aij
ai2
(1)
a ;
(1) 2j
a22
i, j = 3, 4, . . . , n.
(1)
(1)
(1)
(2)
(2)
(2)
(5.14)
a(n1)
xn = b(n1)
,
nn
n
where,
(k1)
(k)
(k1)
aij = aij
aik
(k1)
akk
(k1)
akj
(0)
299
Program 5.5
.
/* Program Gausselimination
Program to find the solution of a system of linear equations by
Gauss elimination method. Partial pivoting is used. */
#include<stdio.h>
#include<stdlib.h>
#include<math.h>
#define zero 0.00001
void main()
{
int i,j,k,n,m;
float a[10][10],b[10],x[10],temp;
printf("\nEnter the size of the coefficient matrix ");
scanf("%d",&n);
printf("Enter the elements rowwise ");
for(i=1;i<=n;i++) for(j=1;j<=n;j++) scanf("%f",&a[i][j]);
printf("Enter the right hand vector ");
for(i=1;i<=n;i++) scanf("%f",&b[i]);
/* augment A with b[i], i.e., a[i][n+1]=b[i] */
m=n+1;
for(i=1;i<=n;i++) a[i][m]=b[i];
for(k=1;k<=n;k++)
{
/* finds pivot element and its position */
temp=fabs(a[k][k]); j=k; /* initial maximum */
for(i=k+1;i<=n;i++)
if(temp<fabs(a[i][k]))
{
temp=fabs(a[i][k]); j=i;
}
if(fabs(a[j][k])<=zero) /* if a[j][k]=0 */
{
printf("The matrix is singular:");
printf("The system has either no solution or many solutions");
exit(0);
}
if(j!=k) /* interchange k and j rows */
{
for(i=1;i<=m;i++)
301
5.6
303
and produces the solution of the system without using the back substitution. At the end
of GaussJordan method the system of equations (5.2) reduces to the following form:
1
0
0
1
x1
x2
..
.
xn
b1
b
2
= .. .
.
(5.15)
bn
..
A.b
(5.16)
1 1 1
3
x1
Solution. Here A = 2 3 1 , x = x2 and b = 6 .
x3
1 1 1
3
..
The augmented matrix A.b is
..
A.b =
2
1
3
1 1
1
1
0 2
.
1 .. 3
1 .. 6
.
1 .. 3
.
1 .. 3
R = R2 2R1 ,
.
2
1 .. 0 R = R3 R1
3
.
2 .. 6
1 1
0 1
0 0
1 1
0 1
0 0
1 0
0 1
0 0
1 0
0 1
0 0
.
1 .. 3
.
1 .. 0 R3 = R3 + 2R2
.
4 .. 6
.
1 .. 3
1
.
1 .. 0 R3 = 4 R3
.
1 .. 3/2
.
2 .. 3
.
1 .. 0 R1 = R1 R2
.
1 .. 3/2
.
0 .. 0
R = R1 2R3 ,
.
1
0 .. 3/2 R = R2 + R3
2
.
1 .. 3/2
5.7
5.7.1
(5.17)
l11 0 0 0
u11 u12
l21 l22 0 0
0 u22
u1n
u2n
u3n
.
..
..
.
.
0 unn
u13
u23
u33
..
.
305
(5.19)
=
=
=
b1
b2
b3
bn .
(5.20)
After determination of z, one can compute the value of x i.e., x1 , x2 , . . . , xn from the
equation Ux = z i.e., from the following equations by the backward substitution.
u11 x1 + u12 x2 + u13 x3 + + u1n xn
u22 x2 + u23 x3 + z2n xn
u33 x3 + u23 x3 + u3n xn
=
=
=
=
=
z1
z2
z3
zn1
zn .
(5.21)
When uii = 1, for i = 1, 2, . . . , n, then the method is known as Crouts decomposition method. When lii = 1, for i = 1, 2, . . . , n then the method is known as
Doolittles method for decomposition. In particular, when lii = uii for i = 1, 2, . . . , n
then the corresponding method is called Choleskys decomposition method.
Procedure to compute L and U
Here,
we assume that uii = 1 for i = 1, 2, . . . , n. From the relation LU =A, i.e., from
l11 u13
l11 u1n
l11 l11 u12
l31 l31 u12 + l32 l31 u13 + l32 u23 + l33 l31 u1n + l32 + u2n + l33 u3n
..
..
..
..
..
.
.
.
.
.
ln1 ln1 u12 + ln2 ln1 u13 + ln2 u23 + ln3 ln1 u1n + ln2 u2n + + lnn
=
a31 a32 a33 ann
we obtain
a1j
li1 = ai1 , i = 1, 2, . . . , n and u1j =
, j = 2, 3, . . . , n.
l11
The second column of L and the second row of U are determined from the relations
li2 = ai2 li1 u12 , for i = 2, 3, . . . , n,
a2j l21 u1j
for j = 3, 4, . . . , n.
u2j =
l22
Next, third column of L and third row of U are determined in a similar way.
In general, lij and uij are given by
lij = aij
aij
uij =
j1
k=1
i1
lik ukj , i j
(5.22)
lik ukj
k=1
, i<j
lii
uii = 1, lij = 0, j > i and uij = 0, i > j.
(5.23)
x=U
z.
(5.24)
(5.25)
It may be noted that the computation of inverse of a triangular matrix is easier than
an arbitrary matrix.
The inverse of A can also be determined from the relation
A1 = U1 L1 .
(5.26)
i=1
307
The inverse of lower (upper) triangular matrix is also a lower (upper) triangular
matrix.
*
*
n
n
Since A = LU, A = LU = LU =
lii
uii .
i=1
i=1
2 2 1
A = 5 1 3
3 4 1
into the form LU, where L and U are lower and upper triangular matrices and hence
solve the system of equations 2x1 2x2 +x3 = 2, 5x1 +x2 3x3 = 0, 3x1 +4x2 +x3 =
9. Determine L1 and U1 and hence nd A1 . Also determine A.
2 2 1
Solution. Let 5 1 3
3 4 1
l11 u13
1 u12 u13
l11 l11 u12
l11 0 0
.
= l21 l22 0 0 1 u23 = l21 l21 u12 + l22 l21 u13 + l22 u23
0 0 1
l31 l31 u12 + l32 l31 u13 + l32 u23 + l33
l31 l32 l33
Comparing both sides, we have
l31 = 3
l11 = 2, l21 = 5
or, u12 = 2/l11 = 1
l11 u12 = 2
or, u13 = 1/l11 = 1/2
l11 u13 = 1
or, l22 = 1 l21 u12 = 6
l21 u12 + l22 = 1
or, l32 = 4 l31 u12 = 7
l31 u12 + l32 = 4
or, u23 = (3 l21 u13 )/l22 = 11/12
l21 u13 + l22 u23 = 3
l31 u13 + l32 u23 + l33 = 1 or, l33 = 1 l31 u13 l32 u23 = 71/12.
Hence L and U are given by
1
2 0 0
1 1
2
.
L = 5 6 0 ,
U = 0 1 11
12
71
3 7 12
0 0
1
Second Part. The given system
of equations
can be
written asAx = b, where
2
2 2 1
x1
A = 5 1 3 , x = x2 , b = 0 .
x3
9
3 4 1
2 0 0
2
y1
5 6 0 y2 = 0 ,
9
y3
3 7 71
12
That is,
2y1
5y1 + 6y2
3y1 + 7y2 +
= 2,
= 0,
71
y3 = 9.
12
1
1
1 1
x1
2
0 1 11 x2 = 5 ,
6
12
x3
2
0 0
1
i.e.,
1
x1 x2 + x3 = 1
2
11
5
x2 x3 =
12
6
x3 = 2.
Solution of this system of equations is
5 11
1
x3 = 2, x2 = +
2 = 1, x1 = 1 + x2 x3 = 1.
6 12
2
Hence the required solution is x1 = 1, x2 = 1, x3 = 2.
Third Part. Applying the GaussJordan method to nd L1 . The necessary augmented matrix is
.
2 0 0 .. 1 0 0
..
.
L.I =
5 6 0 .. 0 1 0
..
3 7 71
12 . 0 0 1
309
.
1 0 0 .. 12 0 0
.. 5
R1 = 1 R1 , R2 = R2 5R1 , R3 = R3 3R1
0
6
0
.
1
0
2
2
71 ..
3
0 7 12 . 2 0 1
.
1
1 0 0 ..
0
0
2
1
.. 5
1
0 1 0 . 12 6 0 R2 = 6 R2 , R3 = R3 7R2
.. 17 7
0 0 71
12 .
12 6 1
.
1
0
0
1 0 0 ..
2
.. 5
.
1
0
0 1 0 . 12
6
.. 17 14 12
0 0 1 . 71 71 71
1/2
0
0
0 .
Hence L1 = 5/12 1/6
17/71 14/71 12/71
The value of U1 can also be determined in a similar way. Here we apply another
method based on the property inverse of a triangular matrix is a triangular matrix
of same shape.
1 b12 b13
Let U1 = 0 1 b23 .
0 0 1
1
Then U U = I gives
1 b12 b13
1 1 1/2
1 0 0
0 1 b23 0 1 11/12 = 0 1 0 .
0 0 1
0 0
1
0 0 1
1 0 0
1 1 + b12 12 11
12 b12 + b13
= 0 1 0.
0
i.e.,
1
11
12 + b23
0 0 1
0
0
1
Equating both sides
1 11
1 11
5
1 + b12 = 0 or, b12 = 1, b12 + b13 = 0 or, b13 = + b12 =
2 12
2 12
12
11
11
+ b23 = 0 or, b23 = .
12
12
Hence
1 1 5/12
U1 = 0 1 11/12 .
0 0 1
A1
1 1 5/12
1/2
0
0
0
= U1 L1 = 0 1 11/12 5/12 1/6
0 0 1
17/71 14/71 12/71
71
1 = 71.
Last Part. A = LU = 2 6
12
Algorithm 5.6 (LU decomposition). This algorithm nds the solution of a system of linear equations using LU decomposition method. Assume that the principal
minors of all order are nonzero.
Algorithm LUdecomposition
Let Ax = b be the systems of equations and A = [aij ], b = (b1 , b2 , . . . , bn )t ,
x = (x1 , x2 , . . . , xn )t .
//Assume that the principal minors of all order are nonzero.//
//Determine the matrices L and U.//
Step 1. Read the matrix A = [aij ], i, j = 1, 2, . . . , n and the right
hand vector b = (b1 , b2 , . . . , bn )t .
1j
for j = 2, 3, . . . , n;
Step 2. li1 = ai1 for i = 1, 2, . . . , n; u1j = al11
uii = 1 for i = 1, 2, . . . , n.
Step 3. For i, j = 2, 3, . . . , n compute the following
j1
lik ukj , i j
lij = aij
k=1
i1
uij = aij
lik ukj /lii , i < j.
k=1
Print x1 , x2 , . . . , xn as solution.
end LUdecomposition
Program 5.6
.
/* Program LUdecomposition
Solution of a system of equations by LU decomposition method.
Assume that all order principal minors are nonzero. */
#include<stdio.h>
void main()
{
float a[10][10],l[10][10],u[10][10],z[10],x[10],b[10];
int i,j,k,n;
printf("\nEnter the size of the coefficient matrix ");
scanf("%d",&n);
printf("Enter the elements rowwise ");
for(i=1;i<=n;i++) for(j=1;j<=n;j++) scanf("%f",&a[i][j]);
printf("Enter the right hand vector ");
for(i=1;i<=n;i++) scanf("%f",&b[i]);
/* computations of L and U matrices */
for(i=1;i<=n;i++) l[i][1]=a[i][1];
for(j=2;j<=n;j++) u[1][j]=a[1][j]/l[1][1];
for(i=1;i<=n;i++) u[i][i]=1;
for(i=2;i<=n;i++)
for(j=2;j<=n;j++)
if(i>=j)
{
l[i][j]=a[i][j];
for(k=1;k<=j1;k++) l[i][j]=l[i][k]*u[k][j];
}
else
{
u[i][j]=a[i][j];
for(k=1;k<=i1;k++) u[i][j]=l[i][k]*u[k][j];
u[i][j]/=l[i][i];
}
printf("\nThe lower triangular matrix L\n");
for(i=1;i<=n;i++)
{
for(j=1;j<=i;j++) printf("%f ",l[i][j]);
printf("\n");
}
printf("\nThe upper triangular matrix U\n");
311
for(i=1;i<=n;i++)
{
for(j=1;j<i;j++) printf("
");
for(j=i;j<=n;j++) printf("%f ",u[i][j]);
printf("\n");
}
/* solve Lz=b by forward substitution */
z[1]=b[1]/l[1][1];
for(i=2;i<=n;i++)
{
z[i]=b[i];
for(j=1;j<=i1;j++) z[i]=l[i][j]*z[j];
z[i]/=l[i][i];
}
/* solve Ux=z by backward substitution */
x[n]=z[n];
for(i=n1;i>=1;i)
{
x[i]=z[i];
for(j=i+1;j<=n;j++) x[i]=u[i][j]*x[j];
}
printf("The solution is ");
for(i=1;i<=n;i++) printf("%f ",x[i]);
} /* main */
A sample of input/output:
Enter the size of the coefficient matrix 3
Enter the elements rowwise
4 2 1
2 5 2
1 2 7
Enter the right hand vector
3 4 5
The lower triangular matrix L
4.000000
2.000000 4.000000
1.000000 2.500000 5.187500
313
5.8
.
Conventionally, the Gauss elimination is applied to the augmented matrix A..b . This
.
method can also be applied to augmented matrix A..I . In this method, the matrix
A(= LU) becomes an upper triangular matrix U and the unit matrix I becomes the
lower triangular matrix, which is the inverse of L. Then the relation AA1 = I becomes
LUA1 = I i.e.,
UA1 = L1 .
(5.27)
The left hand side of (5.27) is a lower triangular matrix and also the matrices U and
L1 are known. Hence by back substitution the matrix A1 can be determined easily.
This method is illustrated by an example below.
Example
5.8.1 Find
the inverse of the matrix A using Gauss elimination method
1 3 4
where A = 1 0 2 .
2 3 1
.
Solution. The augmented matrix A..I is
.
1 3 4 .. 1 0 0
..
.
A.I =
1 0 2 .. 0 1 0
.
2 3 1 .. 0 0 1
1 3 4
R2 R2 R1
0 3 2
R3 R3 + 2R1
0 9 9
..
. 1 0 0
..
. 1 1 0
..
. 2 0 1
..
4
.
1
0
0
R3 R3 + 3R2
..
0
3
2
.
1
1
0
..
0 0 3 . 1 3 1
1
1 3 4
1
1
Here U = 0 3 2 , L = 1
0 0 3
1
1 3 4
x11
0 3 2 x21
x31
0 0 3
0 0
1 0.
3 1
x12 x13
1 0 0
x22 x23 = 1 1 0 .
x32 x33
1 3 1
This implies
x11 + 3x21 + 4x31 = 1
3x21 2x31 = 1
3x31 = 1
1
5
2
These equations give x31 = , x21 = , x11 = .
3
9
3
Again, x12 + 3x22 + 4x32 = 0
3x22 2x32 = 1
3x32 = 3
Solution of this system is x32 = 1, x22 = 1, x12 = 1
x13 + 3x23 + 4x33 = 0
3x23 2x33 = 0
3x33 = 1
1
2
2
Then, x33 = , x23 = , x13 = is the solution of the above equations.
3
9
3
2/3 1 2/3
Hence A1 = 5/9 1 2/9 .
1/3 1 1/3
5.9
Cholesky Method
If the coecient matrix A is symmetric and positive denite then this method is applicable to solve the system Ax = b. This method is also known as squareroot method.
Since A is symmetric then A can be written as
A = LLt ,
(5.28)
(5.29)
315
Let
then
LLt x = b.
(5.30)
Lt x = z
(5.31)
Lz = b.
(5.32)
The vector z can be obtained from (5.32) by forward substitution and the solution
vector x are determined from the equation (5.31) by back substitution. Also, z and x
can be determined by computing the inverse of L only as
z = L1 b and x = (Lt )1 z = (L1 )t z.
(5.33)
l11 0 0 0
l21 l22 0 0 l11 l21 lj1 ln1
0
0
l
A =
j3
n3
l11
2 + l2
l21 l11 l21
lj1 l21 + lj2 l22
22
=
li1 l11 l21 li1 + l22 li2 lj1 li1 + + ljj lij
ln1 l11 l21 ln1 + l22 ln2 lj1 ln1 + + ljj lnj
ln1 l11
.
ln1 li1 + + lni lii
2
2
2
ln1 + ln2 + + lnn
i1
1/2
2 + l2 + + l2 = a
li1
or
l
=
a
lij
, i = 2, 3, . . . , n
ii
ii
ii
i2
ii
j=1
lij = 0, i < j.
(5.34)
for i = j + 1, j + 2, . . . , n
for i = n 2, n 3, . . . , 1; j = i + 1, i + 2, . . . , n 1
1/2
n
2
uik
, i = n 1, n 2, . . . , 1
uii = aii
k=i+1
uij = 0, i > j.
Example 5.9.1 Solve the following system of equations by Cholesky method.
2x1 + x2 x3 = 6
x1 3x2 + 5x3 = 11
x1 + 5x2 + 4x3 = 13.
Solution. The given system of equations is
2 1 1
Ax = b where x = (x1 , x2 , x3 )t , b = (6, 11, 13)t , and A = 1 3 2 .
1 2 4
It is observed
that
A
is
symmetric
and
positive
denite.
l11 0 0
Let L = l21 l22 0 .
l31 l32 l33
2
l11
l11 l21
l11 l31
2 1 1
2 + l2
l21 l31 + l22 l32 = 1 3 2 .
Therefore, LLt = l21 l11 l21
22
2 + l2 + l2
1 2 4
l31 l11 l31 l21 + l32 l22 l31
32
33
Comparing both sides,
we
have
2 = 2 or l
2
l11
11 =
l11 l21 = 1 or l21 = 1/2
l11 l31 = 1 or l31 = 1/ 2
+
1 1/2
2 + l2 = 3 or l
=
l21
22 = (3 2 )
22
5
2
l31 l21 ) = 3/ 10
+
13
2 + l2 + l2 = 4 or l
2 l2 )1/2 =
=
(4
l
l31
33
32
33
31
32
5 .
2
0
0
1.41421
0
0
.
0
Therefore, L = 1/2
5/2 0 = 0.70711 1.58114
0.70711 0.94868 1.61245
1/ 2 3/ 10 13/5
l31 l21 + l32 l22 = 2 or l32 =
1
l22 (2
(5.35)
317
0
1.58114 0.94868 x2 = 5.05963 .
x3
0
0
1.61245
3.22491
i.e.,
5.10
When a matrix is very large and it is not possible to store the entire matrix into the
primary memory of a computer at a time, then matrix partition method is used to
nd the inverse of a matrix. When a few more variables and consequently a few more
equations are added to the original system then also this method is very useful.
Let the coecient matrix A be partitioned as
..
B
.
C
A=
(5.36)
..
D . E
where B is an l l matrix, C is an l m matrix, D is an m l and E is an m m
matrix; and l, m are positive integers with l + m = n.
Let A1 be partitioned as
..
P
.
Q
(5.37)
A1 =
..
R . S
AA1
..
B . C P
=
..
D . E
R
..
. Q I1
=
..
0
. S
..
. 0
,
..
. I2
(5.38)
where I1 and I2 are identity matrices of order l and m respectively. From (5.38), we
have
BP + CR = I1
BQ + CS = 0
DP + ER = 0
DQ + ES = I2 .
Now, BQ + CS = 0 gives Q = B1 CS i.e., DQ = DB1 CS.
Also, from DQ + ES = I2 , we have (E DB1 C)S = I2 .
Therefore, S = (E DB1 C)1 .
Similarly, the other matrices are
Q = B1 CS
R = (E DB1 C)1 DB1 = SDB1
P = B1 (I1 CR) = B1 B1 CR.
It may be noted that, to nd the inverse of A, it is required to determine the inverses
of two matrices B and (E DB1 C) of order l l and m m respectively.
That is, to compute the inverse of the matrix A of order n n, the inverses of two
lower order (roughly half) matrices are to be determined. If the matrices B, C, D, E
are still large to t in the computer memory, then further partition them.
3 3 4
Example 5.10.1 Find the inverse of the matrix A = 2 1 1 using the matrix
1 3 5
partition method. Hence nd the solution of the system of equations
3x1 + 3x2 + 4x3 = 5
2x1 + x2 + x3 = 7
x1 + 3x2 + 5x3 = 6.
3 3 .. 4
..
..
B . C
3 3
2
1
.
1
A=
= , where B = 2 1
.
D .. E
..
1
3 . 5
4
C=
, D= 1 3 , E= 5
1
..
P . Q
and A1 =
, where P, Q, R and S are given by
..
R . S
S = (E DB1 C)1 , R = SDB1 , P = B1 B1 CR, Q = B1 CS.
Now,
B1 =
E DB
C =
S =
R =
P =
=
Q =
1
3
1 3
2 3
1
3
1 3
.
2 3
1 1 3
1
4
5 1 3
= .
2 3
1
3
3
3
1 1 3
3 1 3
= 5 6
2 3
3
1 1 3
1 1 3
4
1
1
5 6
B B CR =
2 3
2 3
1
3
3
2
3
.
9 11
1 1 3
4
1
3=
2 3
1
5
3
Therefore,
A1
Hence,
2
3 1
= 9 11 5 .
5
6 3
2
3 1
5
17
x = A1 b = 9 11 5 7 = 62 .
5
6 3
6
35
319
5.11
(5.39)
a3 x2 + b3 x3 + c3 x4 = d3
an xn1 + bn xn = dn ,
then the coecient matrix is
b1
a2
0
A=
0
0
c1
b2
a3
0
0
0
c2
b3
0
0
0
0
c3
0
0
0
0
0
0
d1
0
0
0
0
d2
0
0
0
0
and d = .
..
(5.40)
It may be noted that the main diagonal and the adjacent coecients on either side of
it consist of only nonzero elements and all other elements are zero. The matrix is called
tridiagonal matrix and the system of equations is called a tridiagonal system. These
type of matrices occur frequently in the solution of ordinary and partial dierential
equations by nite dierence method.
A tridiagonal system can be solved using LU decomposition method.
Let A
= LU where
0
0
1 0 0 0
2 2 0 0
0
0
L =
,
0 0 0 n1 n1 0
0 0 0 0
n n
0
1 1 0 0 0
0 1 2 0 0
0
and U =
.
0 0 0 0 1 n1
0 0 0 0 0
1
Then
1 1
0
0 0
0
1
2 1 2 + 2
2 2
0 0
0
3
2 3 + 3 0 0
0
LU = 0
0
0
0
0 n n n1 + n
321
Now, comparing the matrix LU with A and obtain the nonzero elements of L and
U as
1 = b1 ,
i i = ci ,
i = ai , i = 2, . . . , n
i = bi i1 i = bi ai
i = 1, 2, . . . , n 1
or, i = ci /i ,
ci1
,
i1
i = 2, 3, . . . , n.
ai ci1
, i = 2, 3, . . . , n
i1
i = ai , i = 2, 3, . . . , n
i = bi
(5.41)
(5.42)
i = ci /i , i = 1, 2, . . . , n 1.
The solution of the equation (5.39) i.e., Ax = d where d = (d1 , d2 , . . . , dn )t can be
obtained by solving Lz = d using forward substitution and then solving Ux = z using
back substitution. The solution of Lz = d is given by
z1 =
d1
di ai zi1
, zi =
, i = 2, 3, . . . , n.
b1
i
(5.43)
ci
xi+1 ,
i
i = n 1, n 2, . . . , 1.
c1
= 2 (1).1 = 3
1
1
c2
3 = b3 a3 = 2 3. = 1
2
3
d1
d 2 a2 z 1
= 3,
z2 =
= 3,
z1 =
b1
2
c2
x2 = z2 x3 = 2,
x3 = z3 = 3,
2
2 = b2 a2
d3 a3 z2
=3
3
c1
x1 = z1 x2 = 1.
1
z3 =
(5.44)
c1
=11=0
1
3 = b3 a3
z1 =
x3 =
x2 =
x1 =
z3 =
d 3 a3 z 2
4s 12
=
3
3s 6
323
325
5.12
0
b1 c1 0
a2 b2 c2
0
0 a3 b3
0
T=
0 0 0 an1 bn1 cn1
0 0 0 0
an bn
for any tij = 0 for i j 2.
The entire matrix can be stored using only three vectors c = (c1 , c2 , . . . , cn1 ), a =
(a2 , a3 , . . . , an ), and b = (b1 , b2 , . . . , bn ). We dene a vector d = (d1 , d2 , . . . , dn ) as
,
if i = 1
b1
di = b ai c
(5.45)
i
i1 if i = 2, 3, . . . , n.
di1
If di = 0 for any i n, then, set di = x (x is just a symbolic name) and continue to
compute di+1 , di+2 , . . . , dn in terms of x by using (5.45).
n
di (in general, this is a polynomial in x) evaluated at x = 0 is
The product P =
i=1
the value of T. If P is free from x then the product P directly gives the value of T.
Example 5.12.1 Find the values of the determinants of the following tridiagonal
matrices.
1 1 0
1 1 0
A = 1 1 2 ,
B = 1 2 1 .
0 3 4
0 1 2
Solution. For the matrix A,
a3
4x 6
.
c2 =
d2
x
Therefore, P = d1 d2 d2 = 4x 6, gives A = 6.
d1 = 1, d2 = 0, so, set d2 = x, d3 = b3
1
M.E.A.ElMikkawy, A fast algorithm for evaluating nth order tridiagonal determinants, J. Computational and Applied Mathematics, 166 (2004) 581584.
5.13
The norm of a vector is the size or length of that vector. The norm of a vector x is
denoted by x. This is a real number and satises the following conditions
(i) x 0 and x = 0 i x = 0
(5.46)
(5.47)
(5.48)
n
xi 
(5.49)
i=1
. n
.
xi 2 (Euclidean norm)
(ii) x2 = /
(5.50)
i=1
(5.51)
Let A and B be two matrices such that A + B and AB are dened. The norm of
a matrix A = [aij ] is denoted by A, which satises the following conditions
(i) A 0 and A = 0 i A = 0
(5.52)
(5.53)
(5.54)
(5.55)
(5.56)
(ii) A2 =
327
(5.57)
0
i
(5.58)
(5.59)
j
2 3 4
A = 0 1 5 .
3 2 6
Solution.
A1 = max{2 + 0 + 3, 3 1 + 2, 4 + 5 + 6} = 15
5.14
(5.60)
Note that this system of equations has no solution. But, if we take the approximate
value of 13 as 0.3 then (5.60) becomes
x + 3y = 4
0.3x + y = 1.33.
(5.61)
1
3
(5.62)
x + 3y = 4
0.333x + y = 1.33.
(5.63)
(5.64)
(5.65)
(5.66)
The system (5.65) is called illconditioned when the changes in y are too large compared to those in x. Otherwise, the system is called wellconditioned. If a system is
illconditioned then the corresponding coecient matrix is called an illconditioned
matrix.
1 3
The system (5.62) is illconditioned and the corresponding coecient matrix
0.33 1
is an illconditioned matrix.
Generally, illcondition occurs when A is small. To measure the illcondition of a
matrix, dierent methods are available. One of the useful method is introduced here.
329
The quantity Cond(A), called the condition number of the matrix, dened by
Cond(A) = A A1
(5.67)
where A is any matrix norm, gives a measure of the condition of the matrix
A. The large value of Cond(A) indicates the illcondition of a matrix or the associated
system of equations.
1 3
4 3
Let A =
and B =
be two matrices.
0.33 1
3 5
5 3
100 300
1
1
1
.
Then A =
and B = 11
3 4
33 100
(A) =
A
r1 r2 r n
(5.68)
otherwise A is wellconditioned.
1 3
For the matrix A =
, r1 = 10, r2 = 1.05304, A = 0.01,
0.33 1
0.01
4 3
= 0.003 and for B =
, r1 = 5, r2 = 34,
(A) =
3 5
10 1.05304
11
= 0.37730.
B = 11, (B) =
5 34
Hence A is illconditioned while B is wellconditioned.
5.14.1
Some methods are available to solve an illconditioned system of linear equations. One
straight forward technique is to carry out the calculations with more number of significant digits. But, the calculations with more signicant digits is timeconsuming. One
suggested method is to improve upon the accuracy of the approximate solution by an
iterative method. This iterative method is discussed below.
aij xj = bi , i = 1, 2, . . . , n.
(5.69)
j=1
12 , . . . , x
1n be an approximate solution of (5.69). Since this is an approximate
Let x
11 , x
n
aij x
1j is not necessarily equal to bi . Let bi = 1bi for this approximate solution.
solution,
j=1
aij x
1j = 1bi , i = 1, 2, . . . , n.
(5.70)
j=1
i.e.,
j=1
n
aij (xj x
1j ) = (bi 1bi )
aij i = di
(5.71)
j=1
1i , di = bi 1bi , i = 1, 2, . . . , n.
where i = xi x
Now, the solution for i s is obtained by solving the system (5.71). Hence the new
solution is given by xi = i + x
1i and these values are better approximations to xi s. This
technique can be repeated again to improve the accuracy.
5.15
The conventional matrix inverse (discussed in Section 5.3) is widely used in many areas
of science and engineering. It is also well known that conventional inverses can be
determined only for square nonsingular matrices. But, in many areas of science and
engineering such as statistics, data analysis etc. some kind of weak inverses of singular
square and rectangular matrices are very much essential. The inverses of such types of
matrices are known as generalized inverse or ginverse. A number of works have
been done during last three decades on ginverse. The generalized inverse of an m n
matrix A is a matrix X of size n m. But, dierent types of generalized inverses are
dened by various authors. The following matrix equations are used to classify the
dierent types of generalized inverses for the matrix A:
(i)
(ii)
(iii)
(iv)
AXA = A
XAX = X
AX = (AX)
XA = (XA) , ( denotes the conjugate transpose).
(5.72)
331
(5.73)
t +
(ii) (A ) = (A ) ;
(iii) A
(5.74)
= A , if A is nonsingular
(5.75)
(5.76)
t 1
(5.77)
(5.78)
(5.79)
(5.80)
(5.81)
The Euclidean norm x = x x is minimum for any choice of arbitrary inverse.
3
Leastsquares solution minimizes Ax b for an inconsistent system.
4
For the proof of the algorithm see Greville, T.N.E, The pseudoinverse of a rectangular or singular
matrix and its application to the solution of system of linear equations, SIAM Review, 1 (1959) 3843.
2
(5.82)
A=
= (1 2 . . . k . . . n )
am1 am2 amk amn
a1k
a2k
t
1 t
A+
1 .
1 = (1 1 )
(5.83)
(5.84)
and k = k Ak1 k .
(5.85)
If k = 0, then compute
else
k = k+ = (kt k )1 kt ;
(5.86)
k = (1 +
(5.87)
kt k )1 kt A+
k1 .
A+
k1 k k .
k
(5.88)
inverse)
of
2
1
0
1
2
Solution. Here 1 = 1 , 2 = 0 , 3 = 1 , 4 = 0 , A1 = 1 .
0
1
1
2
0
1
2
t )1 t = 2 1 0 1
2 1 0 = 15 2 1 0 = 25 15 0
=
(
A+
1 1
1
1
0
2 1 1
0 = 2
2 = A+
=
1 2
5 5 0
5
1
1
1
2
5
2 = 2 A1 2 = 0 1 25 = 25 = 0 (the null column vector).
1
0
1
1 1
5
1 2
Hence 2 = 2+ = (2t 2 )1 2t = 15 25 1 25
5 5 1
1
= 56 15 25 1 = 16 13 56
1
2 1
2 2 = 25 16 13 56 = 15
15
3
1 1 1
A1 2 2
+
= 13 31 35
A2 =
2
6
3
6
1 1 1
3
+
3
3
3
Now, 3 = A2 3 = 1 1 5 1 =
7
6 3
6
6
1
1
0
2 1
6
23
= 13 = 0.
3 = 3 A2 3 = 1 1 0
7
6
1
0 1
16
Hence 3 = 3+ = (3t 3 )1 3t = 6 16 13 16 = 1 2 1
2
2 4 2
3
3 3 3
1
2
1
3 3 =
=
7
7
7
7
6
6 3 6
+
1 1 1
A2 3 3
= 1 2 2
A+
3 =
3
1 2 1
1 1 1
1
1
1 2 2 0 = 3 ,
Now, 4 = A+
3 4 =
1 2 1
2
1
0
1
1
2 1 0
1
1
0
0
1
0
1
3
0
= 0.
4 = 4 A3 4 =
0
2
2
0 1 1
1
2
(the null column vector)
333
1 1 1
1
1 2 2 = 5
4 = (1 + 4t 4 )1 4t A+
3 = 12 1 3 1
12
1 2 1
2
3
4 4 =
12 4 3 = 5/4 9/4
5/12 3/4 2/3
1
7/12
1/4
1/3
+
1/4 1/4 0
A3 4 4
+
=
=
A+
4
7/12 5/4 1/3 = A
4
5/12 3/4 2/3
+
The given system
solution isx =
is Ax = b and its
A b.
3 2
4 3
Note 5.15.1 It may be veried that A+ satises all the conditions (5.72). Again,
1
as A3  = 1 i.e., A3 is nonsingular. In addition to this, for this A,
A+
3 = A3
+
AA = I4 , but A+ A = I4 , the unit matrix of order 4.
The ginverse A+ of A can also be computed using the formula (5.80).
5.16
(5.89)
where A, x and b are of order mn, n1 and m1 respectively. Here, we assume that
(5.89) is inconsistent. Since (5.89) is inconsistent, the system has no solution. Again,
since there may be more than one xl (leastsquares solutions) for which Ax b is
minimum, there exist one such xl (say xm ) whose norm is minimum. That is, xm is
called minimum norm least squares solution if
xm xl
(5.90)
for all x.
(5.91)
335
The minimum norm least squares solution can be determined using the relation
x = A+ b.
(5.92)
am1 x1 + am2 x2 + amn xn bm
(5.93)
i=1 j=1
(5.94)
(5.95)
n
m
(aij xj bi )2 .
(5.96)
i=1 j=1
This method is not suitable for a large system of equations, while the method stated
in equation (5.92) is applicable for a large system also.
3 6
2 4
and hence nd
3
6
3
, 2 =
, A1 =
.
Solution. Let 1 =
2
4
2
3 2
t
1 t = 1 3 2 =
A+
1
1 = (1 1 )
13
13 13 ,
3 2 6
2 = A+
= 2,
1 2 = 13 13
4
6
3
0
2 = 2 A1 2 =
.2 =
= 0 (a null vector),
4
2
0
6 4
3 2
= 15 .2. 13
=
2 = (1 + 2t 2 )1 2t A+
1
13
65 65
8
2 2 = 12
65 65
3 2
+
A1 2 2
+
65
= 65
= A+ , which is the ginverse of A.
Therefore, A2 =
6 4
2
65 65
Second Part: The given
equations
can
be written
as
3 6
x
9
Ax = b, where A =
, x=
, A=
.
2 4
y
5
Then the least squares solution
is
given by
3
2
1
9
1
x = A+ b, i.e., x = 65
= 37
65 2 .
6 4
5
74
37
Hence the least squares solution is x = , y = .
65
65
Example 5.16.2 Find the least squares solution of the following equations x + y =
3.0, 2x y = 0.03, x + 3y = 7.03, and 3x + y = 4.97. Also, estimate the residue.
Solution. Let x, y be least squares solution of the given system. Then the square of
residues S is
S = (x + y 3.0)2 + (2x y 0.03)2 + (x + 3y 7.03)2 + (3x + y 4.97)2 .
We choose x and y in such a way that S is minimum. Therefore
S
S
= 0 and
= 0.
x
y
That is,
2(x + y 3.0) + 4(2x y 0.03) + 2(x + 3y 7.03) + 6(3x + y 4.97) = 0
and 2(x + y 3.0) 2(2x y 0.03) + 6(x + 3y 7.03) + 2(3x + y 4.97) = 0.
337
Iteration Methods
If the system of equations has a large number of variables, then the direct methods
are not much suitable. In this case, the approximate numerical methods are used to
determine the variables of the system.
The approximate methods for solving system of linear equations make it possible to
obtain the values of the roots of the system with the specied accuracy as the limit of
the sequence of some vectors. The process of constructing such a sequence is known as
the iterative process.
The eciency of the application of approximate methods depends on the choice of
the initial vector and the rate of convergence of the process.
The following two approximate methods are widely used to solve a system of linear
equations:
(i) method of iteration (Jacobis iteration method), and
(ii) GaussSeidals iteration method.
Before presenting the iteration methods, some terms are introduced to analyse the
methods.
(k)
Let xi , i = 1, 2, . . . , n be the kth (k = 1, 2, . . .) iterated value of the variable xi and
(k) (k)
(k)
x(k) = (x1 , x2 , . . . , xn )t be the solution vector obtained at the kth iteration.
(k)
The sequence {x }, k = 1, 2, . . . is said to converge to a vector x = (x1 , x2 , . . . , xn )t
if for each i (= 1, 2, . . . , n)
(k)
xi
xi as k .
(5.97)
(k)
= i xi .
(5.98)
(k)
t
(k) = (1 , 2 , . . . , (k)
n ) .
(5.99)
(5.100)
(k+1)
(k)
where ei = xi
xi .
An iteration method is said to be of order p 1 if there exists a positive constant A
such that for all k
(k+1) A(k) p .
5.17
(5.101)
(5.102)
an1 x1 + an2 x2 + + ann xn = bn .
Also, we assume that the quantities aii are pivot elements.
The above equations can be written as
1
(b1 a12 x2 a13 x3 a1n xn )
a11
1
(b2 a21 x1 a23 x3 a2n xn )
x2 =
a22
1
(bn an1 x1 an2 x2 an n1 xn1 ).
xn =
ann
x1 =
(0)
(0)
(5.103)
(0)
x1
(1)
(1)
(1)
(5.104)
Again, substituting x1 , x2 , . . . , xn in the right hand side of (5.103) and obtain the
(2) (2)
(2)
second approximation x1 , x2 , . . . , xn .
(k)
339
(k)
In general, if x1 , x2 , . . . , xn be the kth approximate roots then the next approximate roots are given by
1
(k)
(k)
(b1 a12 x2 a13 x3 a1n x(k)
n )
a11
1
(k+1)
(k)
(k)
x2
=
(b2 a21 x1 a23 x3 a2n x(k)
n )
a22
1
(k)
(k)
(k)
(k+1)
=
(bn an1 x1 an2 x2 an n1 xn1 ).
xn
ann
k = 0, 1, 2, . . . .
(k+1)
x1
(5.105)
The iteration process is continued until all the roots converge to the required number
of signicant gures. This iteration method is called Jacobis iteration or simply the
method of iteration.
The Jacobis iteration method surely converges if the coecient matrix is diagonally
dominant.
5.17.1
(5.106)
j=i
(5.107)
j=i
i xi
n
1
(k)
aij j xj
aii j=1
j=i
or
(k+1)
n
1
(k)
=
aij j .
aii j=1
j=i
That is,
(k+1)
i
1
1
(k)
aij  j
aij  (k) .
aii  j=1
aii  j=1
n
j=i
j=i
Let
A = max
i
n
1
aij  .
aii  j=1
j=i
(5.108)
This relation shows that the rate of convergence of GaussJacobis method is linear.
Again,
(k+1) A(k) A2 (k1) Ak+1 (0) .
That is,
(k) Ak (0) .
(5.109)
or,
(k+1)
[by (5.100)]
(5.110)
This relation gives the absolute error at the (k + 1)th iteration in terms of the error
dierence at kth and (k + 1)th iterations.
Example 5.17.1 Solve the following system of linear equations by GaussJacobis
method correct up to four decimal places and calculate the upper bound of absolute
errors.
27x + 6y z = 54
6x + 15y + 2z = 72
x + y + 54z = 110.
Solution. Obviously, the system is diagonally dominant as
6 +  1 < 27,
6 + 2 < 15,
1 + 1 < 54.
341
x
0
2.00000
1.00878
1.24225
1.15183
1.17327
1.16500
1.16697
1.16614
1.16640
1.16632
1.16635
y
0
4.80000
3.72839
4.14167
4.04319
4.08096
4.07191
4.07537
4.07454
4.07488
4.07477
4.07481
z
0
2.03704
1.91111
1.94931
1.93733
1.94083
1.93974
1.94006
1.93996
1.93999
1.93998
1.93998
e(0)
= (3
105 , 4
105 , 0).
(0)
A
e(0) = 5.71 105 .
1A
Step 6. If xi xni  < ( is an error tolerance) for all i, then goto Step 7 else
set xi = xni for all i and goto Step 5.
Step 7. Print xni , i = 1, 2, . . . , n as solution.
end Gauss Jacobi
Program 5.8
.
/*Program Gauss_Jacobi
Solution of a system of linear equations by GaussJacobis iteration
method. Testing of diagonal dominance is also incorporated.*/
#include<stdio.h>
#include<math.h>
#include<stdlib.h>
void main()
{
float a[10][10],b[10],x[10],xn[10],epp=0.00001,sum;
int i,j,n,flag;
printf("Enter number of variables ");
scanf("%d",&n);
printf("\nEnter the coefficients rowwise ");
for(i=1;i<=n;i++)
for(j=1;j<=n;j++) scanf("%f",&a[i][j]);
printf("\nEnter right hand vector ");
for(i=1;i<=n;i++)
scanf("%f",&b[i]);
for(i=1;i<=n;i++) x[i]=0; /* initialize */
343
5.18
(5.111)
345
(0)
(0)
(5.112)
(0)
To solve these equations an initial approximation x2 , x3 , . . . , xn for the variables x2 , x3 , . . . , xn respectively is considered. Substituting these values to the above
(1)
system and get the rst approximate value of x1 , denoted by x1 . Now, substitut(1)
(0) (0)
(0)
(1)
ing x1 for x1 and x3 , x4 , . . . , xn for x3 , x4 , . . . , xn respectively and we nd x2
from second equation of (5.112), the rst approximate value of x2 . Then substituting
(1) (1)
(1)
(0)
(0)
x1 , x2 , . . . , xi1 , xi+1 , . . . , xn for x1 , x2 , . . . , xi1 , xi+1 , . . . , xn to the ith equation of
(1)
x1
That is,
(k+1)
xi
i1
n
1
(k+1)
(k)
=
aij xj
aij xj , i = 1, 2, . . . , n and k = 0, 1, 2, . . . .
bi
aii
j=1
j=i+1
(k+1)
(k)
Example 5.18.1 Solve the following system of equations by GaussSeidals iteration method, correct up to four decimal places.
27x + 6y z = 54
6x + 15y + 2z = 72
x + y + 54z = 110
Solution. The iteration scheme is
1
(54 6y (k) + z (k) )
27
1
= (72 6x(k+1) 2z (k) )
15
1
= (110 x(k+1) y (k+1) ).
54
x(k+1) =
y (k+1)
z (k+1)
Let y = 0, z = 0 be the initial solution. The successive iterations are shown below.
k
0
1
2
3
4
5
6
2.00000
1.18244
1.16735
1.16642
1.16635
1.16634
y
0
4.00000
4.07023
4.07442
4.07477
4.07480
4.07480
z
0
1.92593
1.93977
1.93997
1.93998
1.93998
1.93998
j=1
347
1
(k+1)
(k)
aij  j
+
aij  j 
aii 
j<i
j>i
1
(k+1)
(k)
aij 
+
aij 
aii 
j<i
Ai
(k+1)
j>i
+ (A Ai )
(k)
.
A Ai (k)
.
1 Ai
A Ai
A.
1 Ai
Therefore the above relation reduces to
Since 0 Ai A < 1 then
(k+1) A(k) .
(5.114)
This shows that the rate of convergence of GaussSeidals iteration is also linear. The
successive substitutions give
(k) Ak (0) .
Now, if A < 1 then (k) 0 as k , i.e., the sequence {x(k) } is sure to converge
when A < 1 i.e.,
n
aij < aii  for all i.
j=1
j=i
In other words the sucient condition for GaussSeidals iteration is that the coecient
matrix is diagonally dominant. The absolute error at the (k + 1)th iteration is given by
(k+1)
A
e(k) when A < 1,
1A
as in previous section.
Note 5.18.2 Usually, the GaussSeidals method converges rapidly than the GaussJacobis method. But, this is not always true. There are some examples in which the
GaussJacobis method converges faster than the GaussSeidals method.
1.00000
0.73333
0.71667
0.72139
0.72312
0.72340
0.72341
y
0
0.25000
0.17917
0.15313
0.14898
0.14881
0.14890
0.14893
z
0
0.55000
0.67083
0.68271
0.68165
0.68099
0.68086
0.68085
k
0
1
2
3
4
5
6
7
8
9
10
2.00000
1.20833
1.00347
0.89757
0.97866
0.99964
1.00163
1.00071
1.00016
1.00001
y
0
1.75000
1.39583
1.10243
0.92328
0.95946
0.98951
0.99901
1.00046
1.00028
1.00008
349
z
0
0.31250
0.79688
1.10243
1.07042
1.02081
1.00280
0.99943
0.99953
0.99985
0.99998
x1
(k)
= 1 + 3x2 ,
(k+1)
x2
(k+1)
= 2 x1
(5.116)
For the
k
0
1
2
3
4
5
6
7
scheme (5.116)
x1 x2
0
1
1
4
2
5
7
22 20
59 61
184 182
547 545
(0,2)
Second equ.
? 6
(1,0)
(2,0)
First equ.
(0,1/3)
Figure 5.1: Illustration of GaussSeidals method for the convergent scheme (5.115.)
6
6
(2,0)
(4,0)
?
(0,2)
Second equ.
Figure 5.2: Illustration of GaussSeidals method for the divergent scheme (5.116).
Algorithm 5.9 (GaussSeidals). This algorithm nds the solution of a system
of linear equations by GaussSeidals iteration method. The method will terminate
(k+1)
(k)
xi  < , where is the supplied error tolerance, for all i.
when xi
j>i
for i = 1, 2, . . . , n.
j>i
Step 6. If xi xni  < ( is an error tolerance) for all i then goto Step 7 else
set xi = xni for all i and goto Step 5.
Step 7. Print xni , i = 1, 2, . . . , n as solution.
end Gauss Seidal
Program 5.9
.
/* Program GaussSeidal
Solution of a system of linear equations by GaussSeidals
iteration method. Assume that the coefficient matrix satisfies
the condition of convergence. */
#include<stdio.h>
#include<math.h>
void main()
{
float a[10][10],b[10],x[10],xn[10],epp=0.00001,sum;
int i,j,n,flag;
printf("Enter number of variables ");
scanf("%d",&n);
printf("\nEnter the coefficients rowwise ");
for(i=1;i<=n;i++)
for(j=1;j<=n;j++) scanf("%f",&a[i][j]);
printf("\nEnter right hand vector ");
for(i=1;i<=n;i++)
scanf("%f",&b[i]);
for(i=1;i<=n;i++) x[i]=0; /* initialize */
/* testing of diagonal dominance may be included here
from the program of GaussJacobis method */
351
do
{
for(i=1;i<=n;i++)
{
sum=b[i];
for(j=1;j<=n;j++)
{
if(j<i)
sum=a[i][j]*xn[j];
else if(j>i)
sum=a[i][j]*x[j];
xn[i]=sum/a[i][i];
}
}
flag=0; /* indicates x[i]xn[i]<epp for all i */
for(i=1;i<=n;i++) if(fabs(x[i]xn[i])>epp) flag=1;
if(flag==1) for(i=1;i<=n;i++) x[i]=xn[i]; /* reset x[i] */
}
while(flag==1);
printf("Solution is \n");
for(i=1;i<=n;i++) printf("%8.5f ",xn[i]);
} /* main */
A sample of input/output:
Enter number of variables 3
Enter the coefficients rowwise
3 1 1
2 5 2
2 4 6
Enter right hand vector
7 9 8
Solution is
2.00000 1.00000 0.00000
5.19
353
system of equations
n
aij xj = bi , i = 1, 2, . . . , n.
(5.117)
j=1
Then
n
(k)
aij xj
= bi , i = 1, 2, . . . , n.
j=1
(k)
If ri
ri
= bi
n
(k)
aij xj .
(5.118)
j=1
Now, the solution can be improved successively by reducing the largest residual to
zero at that iteration.
To achieve the fast convergence of the method, the equations are rearranged in such
a way that the largest coecients in the equations appear on the diagonals. Now, the
largest residual (in magnitude) is determined and let it be rp . Then the value of the
rp
.
variable xp be increased by dxp where dxp =
app
In other words, xp is changed to xp + dxp to relax rp , i.e., to reduce rp to zero. Then
the new solution after this iteration becomes
(k) (k)
(k)
(k)
x(k+1) = x1 , x2 , . . . , xp1 , xp + dxp , xp+1 , . . . , xn(k) .
The method is repeated until all the residuals become zero or very small.
Example 5.19.1 Solve the following system of equations
2x1 + x2 + 9x3 = 12,
x1 7x2 + 2x3 = 4
8x1 + x2 x3 = 8,
by relaxation method taking (0, 0, 0) as initial solution.
Solution. We rearrange the equations to get the largest coecients in the diagonals
as
8x1 + x2 x3 = 8
x1 7x2 + 2x3 = 4
2x1 + x2 + 9x3 = 12.
The residuals r1 , r2 , r3 can be computed from the equations
r1 = 8 8x1 x2 + x3
r2 = 4 x1 + 7x2 2x3
r3 = 12 2x1 x2 9x3 .
r1
8
9.333
0.003
1.122
1.505
0.001
0.135
0.192
0
0.019
0.027
0.003
0.001
0
residuals
max
r2
r3
(r1 , r2 , r3 )
4
12
12
6.666 0.003
9.333
7.833 2.331 7.833
0
3.450 3.450
0.766 0.003 1.505
0.954 0.373
0.954
0.002 0.509
0.509
0.112 0.004
0.192
0.136 0.052 0.136
0.003 0.071 0.071
0.013 0.001
0.027
0.016 0.007
0.016
0.002 0.009
0.009
0
0
0
3
1
2
3
1
2
3
1
2
3
1
2
3
increment
dxp
1.333
1.167
1.119
0.383
0.188
0.136
0.057
0.024
0.019
0.008
0.003
0.002
0.001
0
solution
x1
x2
x3
0
0
0
0
0 1.333
1.167 0 1.333
1.167 1.119 1.333
1.167 1.119 0.950
0.979 1.119 0.950
0.979 0.983 0.950
0.979 0.983 1.007
1.003 0.983 1.007
1.003 1.002 1.007
1.003 1.002 0.999
1.000 1.002 0.999
1.000 1.000 0.999
1.000 1.000 1.000
1.000 1.000 1.000
At this stage all the residuals are zero and therefore the solution of the given system
of equations is x1 = 1.000, x2 = 1.000, x3 = 1.000, which is the exact solution of the
equations.
5.20
aij xj +
n
j=i
aij xj = bi .
(5.119)
355
(k+1)
aij xj
j=1
n
(k)
aij xj
= bi .
(5.120)
j=i
i1
(k+1)
aij xj
j=1
(k)
(k+1)
n
(k)
aij xj .
(5.121)
j=i
(k)
aii di
= wri , i = 1, 2, . . . , n,
(5.122)
(k)
aii xi
i1
j=1
(k+1)
aij xj
n
(k)
aij xj
bi ,
(5.123)
j=i
i = 1, 2, . . . , n; k = 0, 1, 2, . . .
6 (0) (0)
(0) 7t
and x1 , x2 , . . . , xn
is the initial solution. The method is repeated until desired
accuracy is achieved.
This method is called the overrelaxation method when 1 < w < 2, and is called
the under relaxation method when 0 < w < 1. When w = 1, the method becomes
GaussSeidals method.
Best relaxation factor wb
The proper choice of wb can speed up the convergence of the system. In a problem,
Carre took wb = 1.9 and found that the convergence is 40 times faster than when w = 1
(GaussSeidals method). He also observed that when w = 1.875 (a minor variation
of 1.9), the convergence is only two times faster than the GaussSeidals method. In
general, the choice of wb is not a simple task. It depends on the spectral radius of the
coecient matrix.
3x1
(k+1)
4x2
(k+1)
4x3
(0)
(0)
(k)
(k)
(k)
(k)
= 3x1 1.01 3x1 + x2 + 2x3 6
(k)
(k+1)
(k)
(k)
= 4x2 1.01 x1
+ 4x2 + 2x3 5
(k)
(k+1)
(k+1)
(k)
= 4x3 1.01 2x1
+ x2
+ 4x3 7 .
(0)
Let x1 = x2 = x3 = 0.
The detail calculations are shown in the following table.
k
0
1
2
3
4
5
6
7
8
9
x1
0
2.02000
1.20116
0.99557
0.98169
0.99312
0.99879
1.00009
1.00013
1.00005
x2
0
1.77255
1.39665
1.09326
1.00422
0.99399
0.99728
0.99942
0.99999
1.00005
x3
0
0.29983
0.80526
0.98064
1.00838
1.00491
1.00125
1.00009
0.99993
0.99997
Program 5.10
.
/* Program GaussSeidal SOR
Solution of a system of linear equations by GaussSeidal
successive overrelaxation (SOR) method. The relaxation factor w
lies between 1 and 2. Assume that the coefficient matrix
satisfies the condition of convergence. */
#include<stdio.h>
#include<math.h>
void main()
{
float a[10][10],b[10],x[10],xn[10],epp=0.00001,sum,w;
int i,j,n,flag;
printf("Enter number of variables ");
scanf("%d",&n);
printf("\nEnter the coefficients rowwise ");
for(i=1;i<=n;i++)
for(j=1;j<=n;j++) scanf("%f",&a[i][j]);
printf("\nEnter right hand vector ");
for(i=1;i<=n;i++) scanf("%f",&b[i]);
printf("Enter the relaxation factor w ");
scanf("%f",&w);
for(i=1;i<=n;i++) x[i]=0; /* initialize */
for(i=1;i<=n;i++) printf(" x[%d] ",i);printf("\n");
do
{
for(i=1;i<=n;i++)
{
sum=b[i]*w+a[i][i]*x[i];
for(j=1;j<=n;j++)
{
if(j<i)
sum=a[i][j]*xn[j]*w;
else if(j>=i)
sum=a[i][j]*x[j]*w;
xn[i]=sum/a[i][i];
}
}
for(i=1;i<=n;i++) printf("%8.5f ",xn[i]);
printf("\n");
357
5.21
The direct and iterative, both the methods have some advantages and also some disadvantages and a choice between them is based on the given system of equations.
(i) The direct method is applicable for all types of problems (when the coecient
determinant is not zero) where as iterative methods are useful only for particular
types of problems.
(ii) The rounding errors may become large particularly for illconditioned systems
while in iterative method the rounding error is small, since it is committed in
359
the last iteration. Thus for illconditioned systems an iterative method is a good
choice.
(iii) In each iteration, the computational eect is large in direct method (it is 2n3 /3
for elimination method) and it is low in iteration method (2n2 in GaussJacobis
and GaussSeidals methods).
(iv) Most of the direct methods are applied on the coecient matrix and for this
purpose, the entire matrix to be stored into primary memory of the computer.
But, the iteration methods are applied in a single equation at a time, and hence
only a single equation is to be stored at a time in primary memory. Thus iterative
methods are ecient then direct method with respect to space.
5.22
Exercise
pivoting.
8 4
0 5
.
1 2
7 10
11 3 1
2 5 5
1 1 1
1 2 0
0 1 2
1 0 5 .
3 5 1
(ii)
(i)
3 8 7
6 8 9
.
4 2 1
5 2 1
2 5 2
1 0 3 .
(i)
(ii)
1 2 7
3 1 6
9. Solve LY = B, UX = Y
A = LU is given by
4 8 4 0
1 5 4 3
1 4 7 2
1 3 0 2
1
0
0
1/4 1
0
=
1/4 2/3
1
1/4 1/3 1/2
0
4 8
0 0 3
0 0 0
1
0 0
4 0
3 3
.
4 4
0 1
20 1 2
8 1 1
3 20 1
2 1 9 .
(i)
(ii)
2 3 20
1 7 2
14. Find the solution of the following tridiagonal system:
2x1 2x2 = 1
x1 + 2x2 3x3 = 2
2x2 + 2x3 4x4 = 1
x3 x4 = 3.
15. Test the following system for illcondition.
10x + 7y + 8z + 7w = 32
7x + 5y + 6z + 5w = 23
8x + 6y + 10z + 9w = 33
7x + 5y + 9z + 10w = 31.
16. Find the ginverses of the following matrices
(i)
2 3
4 6
(ii)
2 3 5
1 1 0 .
3 1 2
1 1 1
2 3 5
361
c2 x3
d3 x3
..
.
an2 xn2
=
=
=
b1
b2
b3
..
.
c4 x4
..
.
+ dn1 xn1 + cn1 xn = bn1
an1 xn1 + dn xn = bn .
363
1 1/2 1/3 1/4 1/5
1
1/2 1/3 1/4 1/5 1/6
0
, b = 0 .
1/3
1/4
1/5
1/6
1/7
A=
1/4 1/5 1/6 1/7 1/8
0
1/5 1/6 1/7 1/8 1/9
0
(ii) Solve Cx = b where
1.0
0.5
0.33333 0.25
0.2
0.5
0.33333 0.25
0.2
0.16667
0.33333
0.25
0.2
0.16667
0.14286
C=
0.25
0.2
0.16667 0.14286 0.125
0.2
0.16667 0.14286 0.125 0.11111,
[Note that the two matrices A and C are dierent.]
, b =
1
0
0
0
0
Chapter 6
Eigenvalue of a Matrix
(6.1)
(6.2)
that is,
a11 a12
a21 a22
an1
an2
a13
a23
an3
a1n
a2n
=0
ann
(6.3)
(6.4)
2 1 1
A = 1 2 1 .
1 1 2
Solution. The characteristic equation of A is A I = 0.
Therefore
2 1
1
1 2 1 = 0.
1
1 2
By direct expansion this gives (1 )2 (4 ) = 0.
Hence the characteristic equation is (1 )2 (4 ) = 0 and the eigenvalues of A are
1, 1, and 4. The two distinct eigenvectors corresponding to two eigenvalues = 1
and 4 are calculated below.
Eigenvector corresponding to the eigenvalue 1.
Let (x1 , x2 , x3 )T be the eigenvector corresponding to 1. Then
0
2 1 1
1
x1
1 2 1 1 x2 = 0 .
x3
0
1
1 2 1
Thus
x1 x2 + x3 = 0
x1 + x2 x3 = 0
x1 x2 + x3 = 0.
The solution of this system of equations is x3 = 0, x1 = x2 . We take x2 = 1. Then
the eigenvector is (1, 1, 0)T .
Let (y1 , y2 , y3 )T be the eigenvector corresponding to = 4.
Then
0
2 4 1
1
y1
1 2 4 1 y2 = 0
y3
0
1
1 2 4
367
j=1
i=1
= ak1
x1
x2
xn
+ ak2
+ + akk + + akn .
xk
xk
xk
x
i
Since 1 for i = 1, 2, . . . , n,
xk
n
akj .
j=1
n
aij  .
j=1
The theorem is also true for columns, as the eigenvalues of AT and A are same.
(6.6)
where
c1 =
n
aii = T r. A, which is the sum of all diagonal elements of A, called the trace.
i=1
aii aij
c2 =
aji ajj ,
i<j
aii aij
aji ajj
c3 =
i<j<k aki akj
, is the sum of all principal minors of order three of A,
6.2
(6.7)
Sn
=
=
1 + 2 + + n = T r A,
21 + 22 + + 2n = T r A2 ,
n1 + n2 + + nn = T r An .
(6.8)
(6.9)
369
Bn1
Bn
=
=
=
=
=
A,
AD1 ,
AD2 ,
ADn2 ,
ADn1 ,
d1 =
T r. B1 ,
1
d2 =
2 T r. B2 ,
1
d3 =
3 T r. B3 ,
1
dn1 = n1
T r. Bn1 ,
1
dn =
T
r.
Bn ,
n
D1 =
D2 =
D3 =
Dn1 =
Dn =
B1 d 1 I
B2 d 2 I
B3 d 3 I
(6.10)
Bn1 dn1 I
Bn d n I
2 1 1
B1 = A = 0 2 3 , d1 = T r. B1 = 2 + 2 + 4 = 8
1 5 4
2 1 1
6 1 1
13 9 5
3 3 6
B2 = A(B1 d1 I) = 0 2 3 0 6 3 =
1 5 4
1 5 4
2 9 2
1
1
d2 = T r. B2 = (13 + 3 2) = 6
2
2
2 1 1
7 9 5
9 0 0
B3 = A(B2 d2 I) = 0 2 3 3 9 6 = 0 9 0
1 5 4
2 9 4
0 0 9
d3 =
1
1
T r. B3 = (9 9 9) = 9.
3
3
Dn1
Dn1
=
.
dn
cn
(6.11)
Step 3: for i = 2, 3, . . . , n do
Compute
(a) Bi = A(Bi1 di1 I)
(b) di = 1i (sum of the diagonal elements of Bi )
Step 4: Compute ci = di for i = 1, 2, . . . , n.
//ci s are the coecients of the polynomial.//
end LeverrierFaddeev.
Program 6.1
.
/* Program LeverrierFaddeev
This program finds the characteristic polynomial of
a square matrix. From which we can determine all the
eigenvalues of the matrix. */
#include<stdio.h>
#include<math.h>
void main()
{
int n,i,j,k,l;
float a[10][10],b[10][10],c[10][10],d[11];
printf("Enter the size of the matrix ");
scanf("%d",&n);
371
0.000000
6.000000
0.000000
2.000000
3.000000
1.000000
2.000000
1.000000
3.000000
0.000000
2.000000
1.000000
0.000000
8.000000
2.000000
The LeverrierFaddeev method may also be used to determine all the eigenvectors.
Suppose the matrices D1 , D2 , . . . , Dn1 and the eigenvalues 1 , 2 , . . . , n are known.
Then the eigenvectors x(i) can be determined using the formula
e2 + + en1 ,
x(i) = in1 e0 + in2 e1 + n3
i
(6.12)
where e0 is a unit vector and e1 , e2 , . . . , en1 are column vectors of the matrices
D1 , D2 , . . . , Dn1 of the same order as e0 .
Example 6.2.2 Use LeverrierFaddeev
methodto nd characteristic equation and
9 1 9
all eigenvectors of the matrix A = 3 1 3 .
7 1 7
Solution.
B1 =
d1 =
D1 =
B2 =
d2 =
D2 =
9 1 9
A = 3 1 3
7 1 7
T r. B1 = 9 1 7 = 1
9 1 9
1
3 1 3 0
B1 d 1 I =
7 1 7
0
9 1 9
8 1
AD1 = 3 1 3 3 2
7 1 7
7 1
1
1
T r. B2 = (6 + 2 4) = 2
2
2
4 2 6
B2 d 2 I = 0 0 0
4 2 6
0 0
8 1 9
1 0 = 3 2 3
0 1
7 1 8
9
6 2 6
3= 0 2 0
8
4 2 4
373
9 1 9
4 2 6
0 0 0
B3 = AD2 = 3 1 3 0 0 0 = 0 0 0
7 1 7
4 2 6
0 0 0
1
1
d3 = T r. B3 = (0 + 0 + 0) = 0.
3
3
Thus c1 = d1 = 1, c2 = d2 = 2, c3 = d3 = 0.
The characteristic equation is 3 2 2 = 0.
The eigenvalues are 1 = 0, 2 = 1, 3 = 2.
1
8
4
Let e0 = 0 , and then e1 = 3 , e2 = 0 .
0
7
4
(e1 , e2 are the rst columns of the matrices D1 , D2 ).
The formula
x(i) = 2i e0 + i e1 + e2 ,
4
for 1 = 0 gives x(1) = 0 .
4
Similarly, for 2 = 1,
1
8
4
3
x(2) = (1)2 0 + (1) 3 + 0 = 3
0
7
4
3
1
8
4
24
6 .
and for 2 = 2, x(3) = 22 0 + 2 3 + 0 =
0
7
4
18
4
3
24
6 .
Thus the eigenvectors are 0 , 3 and
4
3
18
A square matrix B is said to be similar to another square matrix A if there exists
a nonsingular matrix P such that B = P1 AP. The similar matrices have the same
eigenvalues. A square matrix A is said to be diagonalisable if A is similar to a square
diagonal matrix. A square matrix A of order n is diagonalisable i A has n linearly
independent eigenvectors. If P1 AP is a diagonal matrix and P is orthogonal then
A is said to be orthogonally diagonalisable. It can be proved that a matrix A is
orthogonally diagonalisable i A is real and symmetric.
If a matrix is either diagonal or lower triangular or upper triangular then its eigenvalues are the diagonal elements.
6.3
Several methods are available to determine the eigenvalues and eigenvectors of a matrix.
Here, two methods Rutishauser and Power methods are introduced.
6.3.1
Rutishauser method
4 2
Example 6.3.1 Find all the eigenvalues of the matrix
using Rutishauser
1 1
method.
u11 u12
1 0
.
Solution. Let A = A1 =
0 u22
l21 1
u12
4 2
u11
.
That is,
=
u11 l21 u12 l21 + u22
1 1
This gives u11 = 4, u12 = 2,
l21 = 1/4,
u22 =
3/2.
1 0
4 2
, U1 =
.
Therefore, L1 =
1/4 1
0 3/2
4 2
1 0
7/2 2
=
.
Form A2 = U1 L1 =
0 3/2
1/4 1
3/8 3/2
u11 u12
1 0
.
Again, let A2 = L2 U2 =
0
u22
l21 1
u12
7/2
2
u11
.
=
u11 l21 u12 l21 + u22
3/8 3/2
375
1
0
7/2 2
, U2 =
.
Therefore, L2 =
3/28 1
0 12/7
7/2 2
1
0
23/7
2
=
.
Form A3 = U2 L2 =
0 12/7
3/28 1
9/49 12/7
In this way, we nd
3.17391
2
3.10959
2
A4 =
, A5 =
,
0.10208 1.82609
0.06080 1.89041
3.07049
2
3.04591
2
, A7 =
,
A6 =
0.03772 1.92951
0.02402 1.95409
3.04591
2
3.04073
2
, A9 =
A8 =
0.00788 1.96986
0.00512 1.97504
and so on.
The sequence {Ai } converges slowly to an upper triangular matrix and the diagonal
elements converge to the eigenvalues of A. The exact eigenvalues are 3 and 2, which
are approximated by the diagonal elements of A9 .
6.3.2
Power method
(6.13)
2
n
X2 + + c n
Xn .
= 1 c1 X 1 + c 2
1
1
(6.14)
A X =
..
.
Ak X =
Ak+1 X =
2
2
2
n
c1 X 1 + c2
X2 + + c n
Xn .
1
1
..
.
k
k
2
n
k
1 c1 X 1 + c2
X2 + + c n
Xn .
1
1
k+1
k+1
2
n
k+1
c1 X 1 + c2
1
X2 + + c n
Xn .
1
1
21
(6.15)
(6.16)
(6.17)
When k , then
right hand sides of (6.16) and (6.17) tend to k1 c1 X1 and
i < 1 for i = 2, . . . , n. Thus for k , Ak X = k c1 X1 and
k+1
1
1 c1 X1 , since
1
k
k+1 X. It is well known that two
Ak+1 X = k+1
1 c1 X1 . That is, for k , 1 A = A
vectors are equal if their corresponding components are same. That is,
Ak+1 X
r , r = 1, 2, . . . , n.
1 = lim
(6.18)
k
k
A X
r
The symbol (Ak X)r denotes the rth component of the vector Ak X.
If 2  1 , then the term within square bracket of (6.17) tend faster to c1 X1 , i.e.,
the rate of convergence is fast.
To reduce the round o error, the method is carried out by normalizing (reducing the
largest element to unity) the eigenvector at each iteration. Let X0 be a nonnull initial
(arbitrary) vector (nonorthogonal to X1 ) and we compute
Yi+1 = AXi
Yi+1
Xi+1 = (i+1) ,
for i = 0, 1, 2, . . . .
(6.19)
where (i+1) is the largest element in magnitude of Yi+1 and it is the (i + 1)th approximate value of 1 . Then
(Yk+1 )r
,
r = 1, 2, . . . , n.
(6.20)
1 = lim
k (Xk )r
and Xk+1 is the eigenvector corresponding to the eigenvalue 1 .
Note 6.3.1 The initial vector X0 is usually chosen as X0 = (1, 1, , 1)T . But, if the
initial vector X0 is poor, then the formula (6.20) does not give 1 , i.e., the limit of the
(Y
)r
may not exist. If this situation occurs, then the initial vector must be
ratio (Xk+1
k )r
changed.
377
Note 6.3.2 The power method is also used to nd the least eigenvalue of a matrix
A. If X is the eigenvector corresponding to the eigenvalue then AX = X. If A
is nonsingular then A1 exist. Therefore, A1 (AX) = A1 X or, A1 X = 1 X.
This means that if is an eigenvalue of A then 1 is an eigenvalue of A1 and the same
eigenvector X corresponds to the eigenvalue 1/ of the matrix A1 . Thus, if is largest
(in magnitude) eigenvalue of A then 1/ is the least eigenvalue of A1 .
Note 6.3.3 We observed that the coecient Xj in (6.16) goes to zero in proposition
to (j /1 )k and that the speed of convergence is governed by the terms (2 /1 )k . Consequently, the rate of convergence is linear.
Example 6.3.2 Find the largest eigenvalue in magnitude and corresponding eigenvector of the matrix
1 3 2
A = 1 0 2
3 4 5
Solution. Let the initial vector be X0 = (1, 1, 1)T .
The rst iteration
is given
by
1 3 2
1
6
Y1 = AX0 = 1 0 2 1 = 1 .
3 4 5
1
12
0.50000
Y
1
= 0.08333 .
Therefore (1) = 12 and X1 =
12
1.0000
1 3 2
0.50000
2.75
Y2 = AX1 = 1 0 2 0.08333 = 1.5
3 4 5
1.00000
6.83333
0.40244
Y2
= 0.21951 .
(2) = 6.83333, X2 =
6.83333
1.0000
0.40244
3.06098
1 3 2
Y3 = AX2 = 1 0 2 0.21951 = 1.59756
3 4 5
1.00000
7.08537
0.43201
(3) = 7.08537, X3 = 0.22547 .
1.00000
3.10843
0.43185
Y4 = 1.56799 , X4 = 0.21784 , (4) = 7.19793.
7.19793
1.
3.08691
0.43050
Y5 = 1.56950 , X5 = 0.21880 , (5) = 7.16691.
7.16672
1.
3.08691
0.43073
Y6 = 1.56950 , X6 = 0.21900 , (6) = 7.16672.
7.16672
1.
3.08772
0.43075
Y7 = 1.56927 , X7 = 0.21892 , (7) = 7.16818.
7.16818
1.0
3.08752
0.43074
Y8 = 1.56925 , X8 = 0.21893 , (8) = 7.16795.
7.16795
1.0
3.08752
0.43074
Y9 = 1.56926 , X9 = 0.21893 , (9) = 7.16792.
7.16792
1.0
3.08753
0.43074
Y10 = 1.56926 , X10 = 0.21893 , (10) = 7.16794.
7.16794
1.0
The required largest eigenvalue is 7.1679 correct up to four decimal places and the
corresponding eigenvector is
0.43074
0.21893 .
1.00000
Algorithm 6.2 (Power method). This method determines the largest eigenvalue
(in magnitude) and its corresponding eigenvector of a square matrix A.
Algorithm Power Method
Step 1. Read the matrix A = [aij ], i, j = 1, 2, . . . , n.
Step 2. Set initial vector X0 = (1, 1, 1, . . . , 1)T of n components.
Step 3. Find the product Y = AX0 .
Step 4. Find the largest element (in magnitude) of the vector Y and let it be .
Step 5. Divide all the elements of Y by and take it as X1 , i.e., X1 = Y/.
Step 6. //Let X0 = (x01 , x02 , . . . , x0n ) and X1 = (x11 , x12 , . . . , x1n ).//
If xoi x1i  > for at least i then
set X0 = X1 and goto Step 3.
Step 7. Print as largest eigenvalue and corresponding eigenvector X1 of A.
end Power Method
379
A sample of input/output:
Enter the size of the matrix 5
Enter the elements row wise
3 4 5 6 7
0 0 2 1 3
3 4 5 2 3
3 4 2 3 4
0 1 2 0 0
The given matrix is
3.000000 4.000000 5.000000 6.000000 7.000000
0.000000 0.000000 2.000000 1.000000 3.000000
3.000000 4.000000 5.000000 2.000000 3.000000
3.000000 4.000000 2.000000 3.000000 4.000000
0.000000 1.000000 2.000000 0.000000 0.000000
The largest eigenvalue is 10.41317
The corresponding eigenvector is
1.00000 0.20028 0.62435 0.41939 0.13915
6.3.3
6.4
The methods discussed earlier are also applicable for symmetric matrices, but, due to
the special properties of symmetric matrices some ecient methods are developed here.
Among them three commonly used methods, viz., Jacobi, Givens and Householder are
discussed here.
In linear algebra it is established that all the eigenvalues of a real symmetric matrix
are real.
6.4.1
381
Jacobis method
This method is widely used to nd the eigenvalues and eigenvectors of a real symmetric
matrix. Since all the eigenvalues of A are real and there exist a real orthogonal matrix
S such that S1 AS is a diagonal matrix D. As D and A are similar, the diagonal
elements of D are the eigenvalues of A. But, the computation of the matrix S is not a
simple task. It is obtained by a series of orthogonal transformations S1 , S2 , . . . , Sn , . . .
as discussed below.
Let aij  be the largest element among the odiagonal elements of A. Now, we
construct an orthogonal matrix S1 whose elements are dened as
sij = sin , sji = sin , sii = cos , sjj = cos ,
(6.21)
all other odiagonal elements are zero and all other diagonal elements are unity. Thus
S1 is of the form
1
0
..
.
S1 = .
..
..
.
0
0
0
..
.
0 cos sin 0
..
..
..
..
.
.
.
.
0 sin cos 0
..
..
..
..
.
.
.
.
0
0
0
1
0
1
..
.
0
0
..
.
0
0
..
.
(6.22)
where cos , sin , sin and cos are at the positions (i, i), (i, j), (j, i) and (j, j) respectively.
aii aij
be a submatrix of A formed by the elements aii , aij , aji and ajj .
Let A1 =
aji ajj
an orthogonal transformation is applied which is
To reduce A1 to a diagonal matrix,
cos sin
, where is an unknown quantity and it will be selected
dened as S1 =
sin cos
in such a way that A1 becomes diagonal.
Now,
S1
=
=
A1 S1
cos sin
cos sin
aii aij
aji ajj
sin cos
sin cos
(6.24)
This expression gives four values of , but, to get smallest rotation, should lie in
/4 /4. The equation (6.24) is valid for all i, j if aii = ajj . If aii = ajj then
4 , if aij > 0
(6.25)
=
, if aij < 0.
4
1
Thus the odiagonal elements sij and sji of S1 A1 S1 vanish and the diagonal
elements are modied. The rst diagonal matrix is obtained by computing D1 =
S1
1 A1 S1 . In the next step largest odiagonal (in magnitude) element is selected
from the matrix D1 and the above process is repeated to generate another orthogonal
matrix S2 to compute D2 . That is,
1 1
1
A(S1 S2 ).
D2 = S1
2 D1 S2 = S2 (S1 AS1 )S2 = (S1 S2 )
= (S1 S2 Sk )1 A(S1 S2 Sk )
= S1 AS
(6.26)
where S = S1 S2 Sk .
As k , Dk tends to a diagonal matrix. The diagonal elements of Dk are the
eigenvalues and the columns of S are the corresponding eigenvectors.
The method has a drawback. The elements those are transferred to zero during
diagonalisation may not necessarily remain zero during subsequent rotations. The value
of must be veried for its accuracy by checking whether  sin2 + cos2 1  is
suciently small.
Note 6.4.1 It may be noted that for orthogonal matrix S, S1 = ST .
383
Note 6.4.2 It can be shown that the minimum number of rotations required to transform a real symmetric matrix into a diagonal matrix is n(n 1)/2.
Example
1 2
A=2 1
2 2
6.4.1
Find the eigenvalues and eigenvectors of the symmetric matrix
2
2 using Jacobis method.
1
Solution. The largest odiagonal element is 2 at (1, 2), (1, 3) and (2, 3) positions.
4
2a12
= = i.e., = .
The rotational angle is given by tan 2 =
a11 a22
0
4
Thus the orthogonal matrix
S
is
1
cos /4 sin /4 0
1/2 1/ 2 0
S1 = sin /4 cos /4 0 = 1/ 2 1/ 2 0 .
0
0
1
0
0
1
Then the rst rotation yields
1/ 2 1/2 0
1 2 2
1/2 1/ 2 0
1/ 2 1/ 2 0 2 1 2 1/ 2 1/ 2 0
D1 = S1
1 AS1 =
2 2 1
0
0 1
0
0
1
3
0 2.82843
3
0 4/ 2
.
0
1
0
= 0 1 0 =
2.82843
0
1
1
4/ 2 0
The largest odiagonal element of D1 is now 2.82843 situated at (1, 3) position and
hence the rotational angle is
2a13
1
= 0.61548.
= tan1
2
a11 a33
The second orthogonal matrix S2 is
cos 0 sin
0.81650 0 0.57735
.
0 =
0
1
0
S2 = 0 1
sin 0 cos
0.57735 0 0.81650
Then second rotation gives
D1 S2
D2 = S1
2
0.81650 0 0.57735
3
0 2.82843
0.81650 0 0.57735
0
1
0
0
1
0
0
1
0
=
0.577351 0 0.81650
2.82843 0
1
0.57735 0 0.81650
5 0 0
= 0 1 0 .
0 0 1
1/2 1/ 2 0
0.81650 0 0.57735
0
1
0
S = S1 S2 = 1/ 2 1/ 2 0
0.57735 0 0.81650
0
0
1
2 3 1
Solution. Let A = 3 2 2 .
1 2 1
It is a real symmetric matrix and the Jacobis method is applicable.
The largest odiagonal element is at a12 = a21 and it is 3.
2a12
6
Then tan 2 =
= = , and this gives = .
a11 a22
0
4
Thus the orthogonal matrix S1 is
cos /4 sin /4 0
1/2 1/ 2 0
S1 = sin /4 cos /4 0 = 1/ 2 1/ 2 0 .
0
0
1
0
0
1
The rst rotation gives
1/ 2 1/2 0
2 3 1
1/2 1/ 2 0
1/ 2 1/ 2 0 3 2 2 1/ 2 1/ 2 0
= S1
1 AS1 =
1 2 1
0
0 1
0
0
1
D1
385
5
0 3/2
= 0
1
1/ 2 .
3/ 2 1/ 2 1
6/ 2
2a13
= 1.06066. or, = 12 tan1 (0.69883) = 0.407413
=
tan 2 =
a11 a33
51
0.91815 0 0.39624
.
0
1
0
So, the next orthogonal matrix S2 is S2 =
0.39624 0 0.91815
D1 S2
D2 = S1
2
0.91815 0 0.39624
5
0
2.12132
0.91815 0 0.39624
0
1
0
0
1 0.70711
0
1
0
=
0.39624 0 0.91815
2.12132 0.70711
1
0.39624 0 0.91815
5.91548 0.28018
0
= 0.28018 1.0 0.64923 .
0
0.64923 0.08452
The largest odiagonal element of D2 is 0.64923, present at the position (2, 3). Then
2a23
1
= 1.19727 or, = tan1 (1.19727) = 0.43747.
a22 a33
2
1
0
0
Therefore, S3 = 0 0.90583 0.42365 .
0 0.42365 0.90583
2a12
1
tan1
= 0.03510.
2
a11 a22
0.99938 0.03509 0
S4 = 0.03509 0.99938 0 .
0
0
1
5.92439
0
0.11863
0
1.31255 0.00417 .
D4 = S1
4 D3 S4 =
0.11863 0.00417 0.38816
Therefore, =
0.99977 0 0.02141
.
0
1
0
S5 =
0.02141 0 0.99977
5.92693 0.00009
0
0.00009 1.31255 0.00417 .
D5 = S1
5 D4 S5 =
0
0.00417 0.38562
The largest odiagonal element in magnitude is 0.00417 situated at (2, 3) position.
Then
2a23
1
= 0.00246.
= tan1
2
a22 a33
1
0
0
1
0.00246 .
S6 = 0
0 0.00246
1
5.92693 0.00009
0
0.00009 1.31256
.
0
D6 = S1
6 D5 S6 =
0
0
0.38563
This matrix is almost diagonal and hence the eigenvalues are 5.9269, 1.3126 and
0.3856 correct up to four decimal places.
The eigenvectors are the columns of
Step 3. Find the largest odiagonal element (in magnitude) from D = [dij ] and
let it be dij .
Step 4. //Find the rotational angle .//
If dii = djj then
if dij > 0 then = /4 else = /4 endif;
else
2dij
1
1
;
= 2 tan
dii djj
endif;
Step 5. //Compute the matrix S1 = [spq ]//
Set spq = 0 for all p, q = 1, 2, . . . , n
skk = 1, k = 1, 2, . . . , n
and sii = sjj = cos , sij = sin , sji = sin .
Step 6. Find D = ST
1 D S1 and S = S S1 ;
Step 7. Repeat steps 3 to 6 until D becomes diagonal.
Step 8. Diagonal elements of D are the eigenvalues and the columns of S are
the corresponding eigenvectors.
end Jacobi
Program 6.3
.
/* Program Jacobis Method to find eigenvalues
This program finds all the eigenvalues and the corresponding
eigenvectors of a real symmetric matrix. Assume that the
given matrix is real symmetric. */
#include<stdio.h>
#include<math.h>
void main()
{
int n,i,j,p,q,flag;
float a[10][10],d[10][10],s[10][10],s1[10][10],s1t[10][10];
float temp[10][10],theta,zero=1e4,max,pi=3.141592654;
printf("Enter the size of the matrix ");
scanf("%d",&n);
printf("Enter the elements row wise ");
for(i=1;i<=n;i++) for(j=1;j<=n;j++) scanf("%f",&a[i][j]);
printf("The given matrix is\n");
for(i=1;i<=n;i++) /* printing of A */
{
for(j=1;j<=n;j++) printf("%8.5f ",a[i][j]); printf("\n");
}
printf("\n");
387
389
Let
0 0
0 0
b4 0
.. .. ..
. . .
0 0 0 0 bn
a1
b2
A= 0
..
.
b2
a2
b3
..
.
0
b3
a3
..
.
0
0
0
..
.
an
0
a1 b2
b2
a
b
2
3
0
a
b
3
3
pn () =
..
..
..
.
.
.
0
0
0
0 0
0
0 0
0
b4 0
0
= 0.
.. .. ..
..
. . .
.
0 bn an
Expanding by minors, the sequence {pn ()} satises the following equations.
p0 () = 1
p1 () = a1
pk+1 () = (ak+1 )pk () b2k+1 pk1 (), k = 1, 2, . . . , n
(6.27)
391
Example
=
=
=
=
1
3
(2 )p1 () 1p0 () = 2 5 + 5
(1 )p2 () 1p1 () = 3 + 62 9 + 2.
1
0
1
2
3
4
p0
+
+
+
+
+
+
p1
+
+
+
+
0
p2
+
+
+
p3 N ()
+ 0
+ 0
1
0
1
+ 2
Hence all the eigenvalues are 2, 0.26795 and 3.73205, the exact values are 2, 2 3.
6.4.3
Givens method
1 0
0
0 0
0 cos sin 0 0
Let S1 =
0 sin cos 0 0 be the orthogonal matrix, where is unknown.
0 0
0
1 0
0 0
0
0 1
Let B1 be the transformed matrix under orthogonal transformation S1 , Then B1 =
S1
1 AS1 .
The elements of (1, 3) and (3, 1) positions are equal and they are a12 sin +a13 cos .
The angle is now obtained by putting this value to zero.
That is,
a13
or,
tan =
.
(6.28)
a12 sin + a13 cos = 0
a12
This transformation is now considered as a rotation in the (2,3)plane. It may be
noted that this computation is more simple than Jacobis method. The matrix B1 has
the form
B1 =
0 a32 a33 a34 a3n .
an1 an2 an3 an4 ann
393
Then apply the rotation in (2,4)plane to convert a14 and a41 to zero. This would
not eect zeros that have been created earlier. Successive rotations are carried out in
the planes (2, 3), (2, 4), . . . , (2, n) where s are so chosen that the new elements at
the positions (1, 3), (1, 4), . . . , (1, n) vanish. After (n 2) such rotations, all elements
of rst row and column (except rst two) become zero. Then the transformed matrix
Bn2 after (n 2) rotations reduces to the following form:
Bn2
a11
a21
=
0
0
0
a12
a22
a32
an2
0
a23
a33
an3
0
a24
a34
an4
0
a2n
a3n
ann
The second row of Bn2 is taken in the same way as of the rst row. The rotations
are made in the planes (3,4), (3,5), . . . , (3, n). Thus, after (n 2) + (n 3) + + 1
(n 1)(n 2)
rotations the matrix A becomes a tridiagonal matrix B of the form
=
2
a1 b2 0 0 0 0
b2 a2 b3 0 0 0
B=
0 b3 a3 b4 0 0 .
0 0 0 0 bn an
In this process, the previously created zeros are not eected by successive rotations.
The eigenvalues of B and A are same as they are similar matrices.
Example
3
A=
1
1 0
0
S1 = 0 cos sin ,
0 sin cos
1
a13
= , i.e., = 0.32175.
a12
3
1
0
0
Therefore, S1 = 0 0.94868 0.31623 .
0 0.31623 0.94868
where tan =
1
0
0
2 3 1
1
0
0
0 0.94868 0.31623 3 1 2 0 0.94868 0.31623
B = S1
1 AS1 =
0 0.31623 0.94868
1 2 1
0 0.31623 0.94868
2
3.16228 0.00001
2.2 .
= 3.16228 0.40001
0.00001
2.2
0.40001
2
3.1623 0
Let B = 3.1623 0.4 2.2 .
0
2.2 0.4
The Sturm sequence is
p0 ()
p1 ()
p2 ()
p3 ()
=
=
=
=
1
2
(0.4 )p1 () 3.16232 p0 () = 2 1.6 10.8
(0.4 )p2 () 2.22 p1 () = 3 + 22 + 15 14.
4
3
1
0
1
2
4
5
p0
+
+
+
+
+
+
+
+
p1
+
+
+
+
+
0
p2
+
+
p3 N ()
+ 0
1
+ 2
+ 2
+ 2
From this table, it is observed that the eigenvalues are located in the intervals
(4, 3), (0, 1) and (4, 5). Any iterative method may be used to nd them. Using
NewtonRaphson method, we nd the eigenvalues 3.47531, 0.87584 and 4.59947 of
B. These are also the eigenvalues of A.
6.4.4
Householders method
This method is applicable to a real symmetric matrix of order nn. It is more economic
and ecient than the Givens method. Here also a sequence of orthogonal (similarity)
transformations is used on A to get a tridiagonal matrix. Each transformation produces
a complete row of zeros in appropriate positions, without aecting the previous rows.
395
(6.29)
(6.30)
Thus
[using (6.30)].
S1 AS = ST AS = SAS,
(6.31)
(6.32)
(6.33)
(6.34)
Now,
S2 = I 2VVT
1
0
0
0
0 1 2s22 2s2 s3 2s2 s4
=
0 2s2 s3 1 2s23 2s3 s4 .
0 2s2 s4 2s3 s4 1 2s24
(6.35)
That is,
a 12 + a 13 + a 14 = a212 + a213 + a214 = q 2 (say)
2
(6.36)
and
a13 2p1 s3 = 0
(6.37)
a14 2p1 s4 = 0
(6.38)
a12
(6.39)
(6.40)
Thus from (6.39), (6.37) and (6.38) the values of s2 , s3 and s4 are obtained as
1
a13
a14
a12
2
,
s4 =
.
(6.41)
s2 =
1
,
s3 =
2
q
2s2 q
2s2 q
It is noticed that the values of s3 and s4 depend on s2 , so the better accuracy can
be achieved if s2 becomes large. This can be obtained by taking suitable sign in (6.41).
Choosing
a12 sign(a12 )
1
2
1+
.
(6.42)
s2 =
2
q
397
The sign of the square root is irrelevant and positive sign is taken. Hence
s3 =
a13 sign(a12 )
,
2q s2
s4 =
a14 sign(a12 )
.
2q s2
Thus rst transformation generates two zeros in the rst row and rst column. The
second transformation is required to create zeros at the positions (2, 4) and (4, 2).
In the second transformation, let V3 = (0, 0, s3 , s4 )T and the matrix
1 0
0
0
0 1
0
0
(6.43)
S3 =
0 0 1 2s23 2s3 s4 .
0 0 2s3 s4 1 2s24
The values of s3 and s4 are to be computed using the previous technique. The new
matrix A3 = S3 A2 S3 is obtained. The zeros in rst row and rst column remain
unchanged while computing A3 . Thus, A3 becomes to a tridiagonal form in this case.
The application of this method for a general n n matrix is obvious. The elements of
the vector Vk = (0, , 0, sk , sk+1 , , sn ) at the kth transformation are given by
.
. n
akr sign(akr )
1
2
1+
,
r = k + 1, where q = /
a2ki
sk =
2
q
i=k+1
si =
aki sign(akr )
,
2qsk
i = k + 1, . . . , n.
Since the tridiagonal matrix An1 is similar to the original matrix A, they have
identical eigenvalues. The eigenvalues of An1 are computed in the same way as in the
Givens method. Once the eigenvalues become available, the eigenvectors are obtained
by solving the homogeneous system of equations (A I)X = 0.
Example
6.4.5 Use the
A=
1 1 3 1 into the tridiagonal form.
1 1 1 2
Solution. First rotation.
= 0.78868,
s2 = 0.88807,
1+
s2 =
2
3
(1)(1)
= 0.32506,
s3 =
2 3 0.88807
1 (1)
s4 =
= 0.32506.
2 3 0.88807
1
0
0
0
0 0.57734 0.57735 0.57735
=
0 0.57735 0.78867 0.21133 .
0 0.57735 0.21133 0.78867
2
1.73204
0
0
1.73204
5.0
0.21132 0.78867
.
A2 = S2 A1 S2 =
0
0.21132 2.28867
0.5
0
0.78867 0.5
1.71133
Second transformation.
V3 = (0, 0, s3 , s4 )T . q = a223 + a224 = 0.81649,
a23 sign(a23 )
1
2
1+
= 0.62941,
s3 = 0.79335,
s3 =
2
q
a24 sign(a23 )
= 0.60876.
2qs3
V3 = (0, 0, 0.79335, 0.60876)T .
s4 =
S3
A3
1 0
0
0
0 1
0
0
= I 2V3 V3T =
0 0 0.25881 0.96592 .
0 0 0.96592 0.25882
2
1.73204
0
0
1.73204
5.0
0.81648
0
.
= S3 A2 S3 =
0
0.81648
2
0.57731
0
0
0.57731
2
Step 3.3.
Compute vr2 =
akr sign(akr )
1
2 1+
q
aki sign(akr )
2qvr
.
Compute vi =
for i = r + 1, . . . , n.
Step 4. Compute the transformation matrix S = I 2V VT .
Step 5. Compute A = S A S.
Step 6. Set k = k + 1, r = r + 1.
Step 7. Repeat steps 3 to 6 until k n 2.
end Householder
Step 3.4.
Program 6.4
.
/* Program Householder method
This program reduces the given real symmetric matrix
into a real symmetric tridiagonal matrix. Assume that
the given matrix is real symmetric. */
#include<stdio.h>
#include<math.h>
void main()
{
int n,i,j,r=2,k,l,sign;
float a[10][10],v[10],s[10][10],temp[10][10],q;
printf("Enter the size of the matrix ");
scanf("%d",&n);
printf("Enter the elements row wise ");
for(i=1;i<=n;i++) for(j=1;j<=n;j++) scanf("%f",&a[i][j]);
printf("The given matrix is\n");
for(i=1;i<=n;i++) /* printing of A */
{
for(j=1;j<=n;j++) printf("%8.5f ",a[i][j]); printf("\n");
}
for(k=1;k<=n2;k++)
{
q=0;
for(i=k+1;i<=n;i++) q+=a[k][i]*a[k][i];
q=sqrt(q);
for(i=1;i<=r1;i++) v[i]=0;
sign=1; if(a[k][r]<0) sign=1;
v[r]=sqrt(0.5*(1+a[k][r]*sign/q));
for(i=r+1;i<=n;i++) v[i]=a[k][i]*sign/(2*q*v[r]);
/* construction of S */
for(i=1;i<=n;i++) for(j=1;j<=n;j++) s[i][j]=2*v[i]*v[j];
399
for(i=1;i<=n;i++) s[i][i]=1+s[i][i];
for(i=1;i<=n;i++) for(j=1;j<=n;j++)
{
temp[i][j]=0;
for(l=1;l<=n;l++) temp[i][j]+=s[i][l]*a[l][j];
}
for(i=1;i<=n;i++) for(j=1;j<=n;j++)
{
a[i][j]=0;
for(l=1;l<=n;l++) a[i][j]+=temp[i][l]*s[l][j];
}
r++;
} /* end of loop k */
printf("The reduced symmetric tridiagonal matrix is\n");
for(i=1;i<=n;i++)
{
for(j=1;j<=n;j++) printf("%8.5f ",a[i][j]);
printf("\n");
}
}/* main */
A sample of input/output:
Enter the size of the matrix 5
Enter the elements row wise
1 1 2 1 1
1 0 1 3 2
2 1 3 1 1
1 3 1 4 0
1 2 1 0 5
The given matrix is
1.00000 1.00000 2.00000 1.00000
1.00000 0.00000 1.00000 3.00000
2.00000 1.00000 3.00000 1.00000
1.00000 3.00000 1.00000 4.00000
1.00000 2.00000 1.00000 0.00000
1.00000
2.00000
1.00000
0.00000
5.00000
0.00000
0.00000
0.00000
6.5
401
Exercise
2 1 3 4
1 1 1 1
2 5 7
3 2 4 1
1 1 1 1
(a) 6 3 4 , (b)
5 3 2 2 , (c) 1 1 1 1 .
5 2 3
3 3 1 1
1 1 1 1
4. Use the LeverrierFaddeev method to nd the eigenvalues and eigenvectors of the
matrices
1 2 1 2
5 6 3
1 2 3
2 1 2 1
2 3
3 2 3
4 5
(a)
, (b) 0 1 1 .
1 1
1 2 1
6. Use power method to nd the largest and the least (in magnitude) eigenvalues of
the following matrices.
4 1 0
2 1 2
3 1 0
1 2
(a)
, (b) 1 2 1 , (c) 5 3 3 , (d) 1 2 2 .
2 3
0 1 1
1 0 2
0 1 1
7. Use Jacobis method to nd the eigenvalues
of the
following matrices.
4 3 2 1
3 2 1
2 2 6
3 4 3 2
to
3 2 1
2
(a) 2 3 2 , (b) 2
1 2 3
6
nd
the eigenvalues
of the
of the
4 3 2
5 4 3
(a) 3 4 3 , (b) 4 5 4 .
2 3 4
3 4 5
11. Find the eigenvalues of the following
tridiagonal
matrices.
4 1 0 0
2 1 0
3 1 0
1 4 1 0
2 1 1
3 7 2
(a) 6 1 3 , (b) 12 20 6 .
12 2 8
20 31 9
1 2 2 2
2 2 1 2
Chapter 7
Dierentiation
f n+1 ()
f n+1 ()
= w(x)
(n + 1)!
(n + 1)!
f n+2 ()
f n+1 ()
+ w(x)
(x).
(n + 1)!
(n + 1)!
(7.1)
The bound of the second term is unknown due to the presence of the unknown
quantity (x).
403
f n+1 (i )
,
(n + 1)!
(7.2)
where min{x, x0 , . . . , xn } < i < max{x, x0 , . . . , xn }. The error can also be expressed
in terms of divided dierence.
f n+1 ()
.
Let E(x) = w(x)f [x, x0 , x1 , . . . , xn ] where f [x, x0 , x1 , . . . , xn ] =
(n + 1)!
Then E (x) = w (x)f [x, x0 , x1 , . . . , xn ] + w(x)f [x, x, x0 , x1 , . . . , xn ].
Now, this expression is dierentiated (k 1) times by Leibnitzs theorem.
k
E (x) =
k
Cr w(i) (x)
Cr w (x)(k i)!f [x, x, . . . , x, x0 , . . . , xn ]
i=0
k
dk1
(f [x, x0 , . . . , xn ])
dxki
ki+1
(i)
i=0
k
k!
i=0
i!
ki+1
w(i) (x)(k i)!f [x, x, . . . , x, x0 , . . . , xn ],
(7.3)
7.2
(x) = y0 + uy0 +
with error
E(x) =
405
1
24u 36 4
60u2 240u + 210 5
3
y0 +
y0 +
(7.7)
(x) = 3 y0 +
h
4!
5!
and so on.
It may be noted that y0 , 2 y0 , 3 y0 , are constants.
The above equations give the approximate derivative of f (x) at arbitrary point x (=
x0 + uh).
When x = x0 , u = 0, the above formulae become
1
1
1
1
1
(7.8)
(x0 ) =
y0 2 y0 + 3 y0 4 y0 + 5 y0
h
2
3
4
5
1
11
5
(7.9)
(x0 ) = 2 2 y0 3 y0 + 4 y0 5 y0 +
h
12
6
1
3
7
(7.10)
(x0 ) = 3 3 y0 4 y0 + 5 y0
h
2
4
and so on.
Error in dierentiation formula based on Newtons forward interpolation
polynomial
The error in Newtons forward interpolation formula is
E(x) = u(u 1) (u n)hn+1
f n+1 ()
.
(n + 1)!
E (x) = hn+1
E (x0 ) = hn
(7.12)
2 y
3 y
3.625
2.0 7.000
3.000
6.625
2.5 13.625
0.750
3.750
10.375
3.0 24.000
0.750
4.500
14.875
3.5 38.875
5.250
20.125
4.0 59.000
0.750
407
7.3
(x) = yn + vyn +
and so on.
The above formulae are used to determine the approximate dierentiation of rst,
second and third, etc. order at any point x where x = xn + vh.
(7.16)
(7.17)
(7.18)
f n+1 ()
,
(n + 1)!
x xn
and min{x, x0 , x1 , . . . , xn } < < max{x, x0 , x1 , . . . , xn }.
h
f n+1 ()
d
[v(v + 1)(v + 2) (v + n)]
dv
(n + 1)!
v(v + 1)(v + 2) (v + n) n+2
(1 ),
f
+hn+1
(n + 1)!
E (x) = hn
E (xn ) = hn
(7.19)
409
x
5
8
12
17
26
x 2 x 3 x 4 x
3
4
5
9
1
1
4
0
3
The velocity is
1
1
1
1
dx
=
xn + 2 xn + 3 xn + 4 xn +
dt
h
2
3
4
1
1
1
1
= 9+ 4+ 3+ 3
1
2
3
4
= 12.75.
The acceleration is
d2 x
1 2
11 4
3
=
x
+
x
+
x
+
n
n
n
dt2
h2
12
1
11
= 2 4+3+
3 = 9.75.
1
12
Example 7.3.2 A slider in a machine moves along a xed straight rod. Its distance
x cm along the rod are given in the following table for various values of the time t
(in second).
t (sec) : 1.0
1.1
1.2
1.3
1.4
1.5
x (cm) : 16.40 19.01 21.96 25.29 29.03 33.21
Find the velocity and the acceleration of the slider at time t = 1.5.
Solution. The backward dierence table is
t
1.0
1.1
1.2
1.3
1.4
1.5
Here h = 0.1.
x
16.40
19.01
21.96
25.29
29.03
33.21
x 2 x 3 x 4 x 5 x
2.61
2.95
3.33
3.74
4.18
0.34
0.38 0.04
0.41 0.03 0.01
0.44 0.03 0.00 0.01
1
1 2 1 3 1 4 1 5
dx
=
+ + + + + xn
dt
h
2
3
4
5
1
1
1
1
1
4.18 + 0.44 + 0.03 + 0.00 + 0.01
=
0.1
2
3
4
5
= 44.12.
1
11 5
d2 x
2
3
= 2 + + + xn
dt2
h
12
11
5
1
0.44
+
0.03
+
0.00
+
0.01
=
(0.1)2
12
6
= 47.83.
Hence velocity and acceleration are respectively 44.12 cm/sec and 47.83 cm/sec2 .
7.4
+
2 y1 +
1!
2
2!
3!
2
5 5u3 + 4u 5 y
5y
u4 u2
u
+
3
2
+
4 y2 +
+
4!
5!
2
x x0
.
where u =
h
(x) = y0 +
h
2
6
2
3
4
2
5
5
2u u 4
5u 15u + 4 y3 + y2
+
y2 +
+ .
12
120
2
3
3
2
1
y2 + y1 6u 1
(x) = 2 2 y1 + u
+
4 y2
h
2
12
2u3 3u 5 y3 + 5 y2
+ .
+
12
2
(7.20)
(7.21)
(7.22)
At x = x0 , u = 0. Then
1 5 y2 + 5 y3
1 y0 + y1 1 3 y1 + 3 y2
+
+ . (7.23)
(x0 ) =
h
2
6
2
30
2
and
1
1 4
2
(x0 ) = 2 y1 y2 + .
h
12
411
(7.24)
f 2n+1 ()
,
(2n + 1)!
d f
()
+h2n+1 [u(u2 12 )(u2 22 ) (u2 n2 )]
dx (2n + 1)!
f 2n+1 ()
d
= h2n [u(u2 12 )(u2 22 ) (u2 n2 )]
du
(2n + 1)!
f 2n+2 (1 )
,
+h2n+1 [u(u2 12 )(u2 22 ) (u2 n2 )]
(2n + 1)!
E (x) =
(7.25)
(7.26)
Example 7.4.1 Compute the values of (i) f (3), (ii) f (3), (iii) f (3.1), (iv) f (3.1)
using the following table.
x :
1
2
3
4
5
f (x) : 0.0000 1.3863 3.2958 5.5452 8.0472
2 y
3 y
4 y
1.3863
x1 = 2
1.3863
0.5232
1.9095
x0 = 3
3.2958
0.1833
0.3399
2.2494
x1 = 4
5.5452
x2 = 5
8.0472
0.0960
0.0873
0.2526
2.5020
Since x = 3 and x = 3.1 are the middle of the table, so the formula based on central
dierence may be used. Here Stirlings formula is used to nd the derivatives.
(i) Here x0 = 3, h = 1, u = 0.
Then
1 y1 + y0 3 y2 + 3 y1
+
f (3) =
h
2
12
= 2.1020.
=
1
2
12
1
1
1 4
1
2
(ii) f (3) = 2 y1 y2 + = 2 0.3399
0.0960 = 0.3319.
h
12
1
12
= 0.1. Then
(iii) Let x0 = 3, h = 1, u = 3.13
1
1 y1 + y0
3u2 1 3 y2 + 3 y1
f (3.1) =
+ u2 y1 +
h
2
6
2
3
2u u 4
+
y2 +
12
3 (0.1)2 1 0.1833 0.0873
1 1.9095 + 2.2494
+ 0.1 0.3399 +
=
1
2
6
2
3
2 (0.1) 0.1
0.0960 = 2.1345.
+
12
(iv)
1
3 y2 + 3 y1 6u2 1 4
2
f (3.1) = 2 y1 + u
+
y2 +
h
2
12
0.1833 0.0873 6(0.1)2 1
1
+
0.0960
= 2 0.3399 + 0.1
1
2
12
= 0.31885.
7.5
413
n
i=0
yi
, where w(x) = (x x0 )(x x1 ) (x xn ).
(x xi )w (xi )
Then
(x) = w (x)
n
i=0
yi
yi
w(x)
.
(x xi )w (xi )
(x xi )2 w (xi )
n
(7.27)
i=0
n
yi
yi
2w
(x)
(x xi )w (xi )
(x xi )2 w (xi )
n
i=0
n
+2w(x)
i=0
i=0
yi
.
(x xi )3 w (xi )
(7.28)
The formulae (7.27) and (7.28) are valid for all x except x = xi , i = 0, 1, . . . , n.
To nd the derivatives at the points x0 , x1 , . . . , xn , the Lagranges polynomial is
rearranged as
(x) = w(x)
n
i=0
i=j
yi
(x xi )w (xi )
Therefore,
(xj ) = w (xj )
n
i=0
i=j
yj
yi
,
+
(xj xi )w (xi )
xj xi
i=0
n
i=j
(7.29)
n
i=0
i=j
1
.
xj xi
f n+1 ()
, where w(x) = (x x0 )(x x1 ) (x xn ).
(n + 1)!
Thus
E (x) = w (x)
f n+2 (1 )
f n+1 ()
+ w(x)
,
(n + 1)!
(n + 1)!
(7.30)
f n+1 ()
.
(n + 1)!
(7.31)
y2
y3
y1
+
+
= w (2)
(2 3)w (3) (2 5)w (5) (2 6)w (6)
1
1
1
+
+
.
+y0
23 25 26
415
34
136
229
1 1
+
+
+ 13 1
= 14.
f (2) 12
(1) 6 (3) (6) (4) 12
3 4
Also
f (2.5)
w (2.5)
3
i=0
yi
yi
w(2.5)
(2.5 xi )w (xi )
(2.5 xi )2 w (xi )
3
i=0
y1
y2
y3
y0
+
+
+
= w (2.5)
(2.5 2)w (2) (2.5 3)w (3) (2.5 5)w (5) (2.5 6)w (6)
y0
y1
y2
y3
w(2.5)
+
+
+
(2.52)2 w (2) (2.53)2 w (3) (2.55)2 w (5) (2.56)2 w (6)
Now, w (2.5) = 1.5, w(2.5) = 2.1875.
Therefore,
34
136
229
13
+
+
+
f (2.5) 1.5
0.5 (12) (0.5) 6 (2.5) (6) (3.5) 12
34
136
229
13
+
+
+
+ 2.1875
(0.5)2(12) (0.5)26 (2.5)2(6) (3.5)212
= 20.75.
Algorithm 7.1 (Derivative). This algorithm determines the rst order derivative
of a function given in tabular form (xi , yi ), i = 0, 1, 2, . . . , n, at a given point xg, xg
may or may not be equal to the given nodes xi , based on Lagranges interpolation.
Algorithm Derivative Lagrange
Read xi , yi , i = 0, 1, 2, . . . , n.
Read xg; //the point at which the derivative is to be evaluated.//
Compute w (xj ), j = 0, 1, 2, . . . , n, using the function wd(j).
Set sum1 = sum2 = 0;
Check xg is equal to given nodes xi .
If xg is not equal to any node then
for i = 0 to n do
Compute t = yi /((xg xi ) wd(i));
Compute sum1 = sum1 + t;
Compute sum2 = sum2 + t/(xg xi );
417
(7.32)
(7.33)
(7.34)
(7.35)
1 y1+y0 3 y2+3 y1 5 y3+5 y2
+
+ y0
f (x0 )
h
2
12
60
1
1 4
1 6
2
f (x0 ) 2 y1 y2 + y3 y0
h
12
90
7.6
419
(7.36)
(7.37)
Only the rst term of (7.32), gives a simple formula for the rst order derivative
f (xi )
yi+1 yi
y(xi + h) y(xi )
yi
=
=
.
h
h
h
(7.38)
yi yi1
y(xi ) y(xi h)
yi
=
=
.
h
h
h
(7.39)
Adding equations (7.38) and (7.39), we obtain the central dierence formula for rst
order derivative, as
f (xi )
y(xi + h) y(xi h)
.
2h
(7.40)
2 yi
yi+2 2yi+1 + yi
y(xi +2h) 2y(xi + h)+y(xi )
=
=
.
2
2
h
h
h2
(7.41)
2 yi
yi 2yi1 +yi2
y(xi )2y(xi h)+y(xi 2h)
=
=
.
2
2
h
h
h2
(7.42)
2 y1
y1 2y0 + y1
y(x0 + h) 2y(x0 ) + y(x0 h)
=
=
.
2
2
h
h
h2
In general,
f (xi )
(7.43)
Equations (7.41)(7.43) give the threepoint formulae for second order derivative.
The truncation error of the twopoint formula (7.40) is O(h2 ). Assume that f
C 3 [a, b] (i.e., f is continuously dierentiable up to third order within [a, b]) and x
h, x, x + h [a, b]. Then by Taylors series
h2
f (xi ) +
2!
h2
and f (xi h) = f (xi ) hf (xi ) + f (xi )
2!
f (xi + h) = f (xi ) + hf (xi ) +
h3
f (1 )
3!
h3
f (2 )
3!
By subtraction
f (xi + h) f (xi h) = 2hf (xi ) +
f (1 ) + f (2 ) 3
h .
3!
(7.44)
Since f is continuous, by the intermediate value theorem there exist a number so
that
f (1 ) + f (2 )
= f ().
2
Thus, after rearrangement the equation (7.44) becomes
f (xi ) =
.
2h
3!
(7.45)
It may be noted that the rst term of right hand side is twopoint formula while
second term is the truncation error and it is of O(h2 ).
To nd the computers roundo error, it is assumed that f (x0 h) = y(x0 h)+1
and f (x0 + h) = y(x0 + h) + 1 where y(x0 h) and y(x0 + h) are the approximate
values of the original function f at the points (x0 h) and (x0 + h) respectively and
1 and 1 are the roundo errors.
Thus
f (xi ) =
y(xi + h) y(xi h)
+ Etrunc
2h
and
y(xi + h) y(xi h)
+ Etrunc + Eround
2h
y(xi + h) y(xi h)
+E
=
2h
f (xi ) =
where
E = Eround + Etrunc =
1 1 h2 f ()
2h
6
(7.46)
421
is the total error accumulating the roundo error (Eround ) and the truncation error
(Etrunc ).
Let 1  , 1  and M3 = max f (x).
axb
Then from (7.46), the upper bound of the total error is given by
E
M3 h2
1  + 1  h2
+ f () +
.
2h
6
h
6
3
M3
(7.47)
dE
hM3
= 0 i.e., 2 +
= 0. Thus the
dh
h
3
1/3
(7.48)
3
E =
M3
1/3
M3
+
6
3
M3
2/3
.
f (x + 2h) + 8f (x + h) 8f (x h) + f (x 2h) h4 f v ()
+
,
12h
30
(7.49)
where lies between x 2h and x + 2h. Determine the optimal value of h when
(i) Eround  = Etrunc ,
(ii) total error Eround  + Etrunc  is minimum.
Solution. By Taylors series expansion for step length h and h,
f (x + h) = f (x) + hf (x) +
h2
h3
h4
h5
f (x) + f (x) + f iv (x) + f v (1 )
2!
3!
4!
5!
f (x h) = f (x) hf (x) +
h2
h3
h4
h5
f (x) f (x) + f iv (x) f v (2 ).
2!
3!
4!
5!
and
Then by subtraction
f (x + h) f (x h) = 2hf (x) +
(7.50)
The rst term on the right hand side is a fourpoint formula to nd f (x) and second
term is the corresponding truncation error.
Let f (x+2h) = y2 +2 , f (x+h) = y1 +1 , f (xh) = y1 +1 and f (x2h) = y2 +
2 , where yi and i are the approximate values of f (x + ih) and the corresponding
roundo errors respectively. Also let
max
f v ().
M5 =
x2hx+2h
f (x) =
M5 h4
.
30
18
3
2  + 81  + 81  + 2 
=
12h
12h
2h
423
3
45
M5 h4
.
=
or, h4 =
2h
30
M5
1/4
45
and
Thus the optimum value of h is
M5
9 1/4
3 M5 1/4
= (M5 3 )1/4
.
Eround  = Etrunc  =
2 45
80
(i) If Eround  = Etrunc  then
3
4M5 h3
45
= 0, i.e., h5 =
+
.
2
2h
30
4M5
45
4M5
dE
=0
dh
1/5
.
(7.51)
Find the value of f (0.6) using the two and fourpoint formulae
f (x0 + h) f (x0 h)
2h
f (x0 + 2h) + 8f (x0 + h) 8f (x0 h) + f (x0 2h)
and
f (x0 ) =
12h
with step size h = 0.1.
f (x0 ) =
f (0.6)
7.7
The improvement of derivative of a function can be done using this method. This
method reduces the number of function evaluation to achieve the higher order accuracy.
The formula to nd the rst order derivative using two points is
f (x) =
f (x + h) f (x h)
+ Etrunc = g(h) + Etrunc
2h
where Etrunc is the truncation error and g(h) is the approximate rst order derivative
of f (x).
Using Taylors series expansion, it can be shown that, Etrunc is of the following form.
Etrunc = c1 h2 + c2 h4 + c3 h6 + .
The Richardsons extrapolation method combines two values of f (x) obtained
by a certain method with two dierent step sizes, say, h1 and h2 . Generally, h1 and h2
are taken as h and h/2. Thus
f (x) = g(h) + Etrunc = g(h) + c1 h2 + c2 h4 + c3 h6 +
h2
h4
h6
and f (x) = g(h/2) + c1 + c2 + c3 + .
4
16
32
(7.52)
(7.53)
f (x) =
(7.54)
Denoting
4g(h/2) g(h)
3
by
g1 (h/2),
(7.55)
(7.56)
425
This equation shows that g1 (h/2) is an approximate value of f (x) with fourthorder
accuracy. Thus a result accurate up to fourth order is obtained by combining two results
accurate up to second order.
Now, by repeating the above result one can obtain
f (x) = g1 (h/2) + d1 h4 + O(h6 )
h4
f (x) = g1 (h/22 ) + d1 + O(h6 )
16
(7.57)
(7.58)
where
g2 (h/22 ) =
42 g1 (h/22 ) g1 (h/2)
.
42 1
(7.59)
second
order
g(h)
fourth
order
sixth
order
eight
order
g1 (h/2)
g2 (h/22 )
h/2 g(h/2)
g1
h/22
(h/22 )
g(h/22 )
g2
g1
h/23 g(h/23 )
g3 (h/23 )
(h/23 )
(h/23 )
g(h/2) g(h)
4g(h/2) g(h)
= g(h/2) +
.
3
3
Here g(h/2) is more accurate than g(h) and then g1 (h/2) gives an improved approximation over g(h/2). If g(h) < g(h/2), g1 (h/2) > g(h/2) and if g(h/2) < g(h),
g1 (h/2) < g(h/2). Thus the value of g1 (h/2) lies outside the interval [g(h), g(h/2)] or
[g(h/2), g(h)] as the case may be. Thus g1 (h/2) is obtained from g(h) and g(h/2) by
means of an extrapolation operation. So, this process is called (Richardson) extrapolation.
Example 7.7.1 Use Richardsons extrapolation method to nd f (0.5) where
f (x) = 1/x starting with h = 0.2.
Solution. Here h = 0.2 and x = 0.5.
1
1
Then
f (x + h) f (x h)
= 0.5 + 0.2 0.5 0.2 = 4.76190,
g(h) =
2h
2 0.2
1
1
f (x + h/2) f (x h/2)
g(h/2) =
= 0.5 + 0.1 0.5 0.1 = 4.16667.
2(h/2)
0.2
4g(h/2) g(h)
4 (4.16667)(4.76190)
=
= 3.96826.
41
3
Halving the step size further, we compute
Then g1 (h/2) =
1
1
16 (3.99831) (3.96826)
42 g1 (h/22 ) g1 (h/2)
=
2
4 1
15
= 4.00031.
g2 (h/22 ) =
427
g1
g2
3.96826
0.1 4.16667
4.00031
3.99831
0.05 4.04040
Thus, after two steps we found that f (0.5) = 4.00031 while the exact value is
1
= 4.0.
f (0.5) = 2
x x=0.5
Example 7.7.2 For the following table
x : 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
f (x) : 1 2/3 1/2 2/5 1/3 2/7 1/4 2/9 1/5 2/11 1/6
nd the value of f (3) using Richardsons extrapolation.
Solution. Let x = 3 and h = 2.
1 1
f (5) f (1)
f (x + h) f (x h)
=
= 6 2 = 0.083333.
g(h) =
2h
22
4
1 1
f (x + h/2) f (x h/2)
f (4) f (2)
5
3 = 0.066666.
g(h/2) =
=
=
2(h/2)
2
2
4 (0.06666) (0.083333)
4g(h/2) g(h)
=
= 0.061110.
g1 (h/2) =
41
3
2 2
f (3.5) f (2.5)
2
9
7 = 0.063492.
g(h/2 ) = g(0.5) =
=
2 0.5
1
4 (0.063492) (0.066666)
4g(h/22 ) g(h/2)
=
= 0.062434.
g1 (h/22 ) =
41
3
Thus
42 g1 (h/22 ) g1 (h/2)
42 1
16 (0.062434) (0.061110)
= 0.062522.
=
15
g2 (h/22 ) =
f (x + h) f (x h)
;
2h
f (x + h) f (x h)
;
2h
for k = 1 to j do
4k gn (k 1) go (k 1)
;
4k 1
if gn (j 1) gn (j) < then
Print gn (j) as the value of derivative;
Stop;
else
for k = 0 to j do
g0 (k) = gn (k); //set new values as old values//
j = j + 1;
goto 10;
endif;
end Richardson extrapolation
gn (k) =
429
A sample of input/output:
Enter the value of x 1.5
Enter the value of h 0.5
The derivative is 0.59259 at
7.8
1.50000
The cubic spline may be used to determine the rst and second derivatives of a function.
This method works into two stages. In rst stage the cubic splines will be constructed
with suitable intervals and in second stage the rst and second derivatives are to be
determined from the appropriate cubic spline. This method is labourious than the
other methods, but, once a cubic spline is constructed then the method becomes very
ecient. The process of nding derivative is illustrated by as example in the following.
Example 7.8.1 Let y = f (x) = cos x, 0 x /2 be the function. Find the natural cubic spline in the intervals 0 x /4 and /4 x /2 and hence determine
the approximate values of f (/8) and f (/8). Also, use twopoint formula to nd
the value of f (/8). Find the error in each case.
6
[y0 2y1 + y2 ]
h2
96
24
That is, 4M1 = 2 (1 2) or, M1 = 2 (1 2) = 1.007246602.
4 x3
1
2
M1 1 + M1 x +
,
6
4
2 96
1
2
4 (/2 x)3
M1 M1 (/2 x) .
s2 (x) =
6
2 96
431
Twopoint formula.
Let h = /30.
f (/8 + /30) f (/8 /30)
= 0.381984382.
Then f (/8)
2./30
The actual value of f (/8) is sin /8 = 0.382683432.
Therefore, error in cubic spline method is 0.042722272 while in twopoint formula
that is 0.000699050365.
7.9
It is known that, if a function is dierentiable then the maximum and minimum value
of that function can be determined by equating the rst derivative to zero and solving
for the variable. The same method is applicable for the tabulated function.
Now, consider the Newtons forward dierence interpolation formula.
y = f (x) = y0 + uy0 +
where u =
Then
u(u 1) 2
u(u 1)(u 2) 3
y0 +
y0 + ,
2!
3!
(7.61)
x x0
.
h
2u 1 2
3u2 6u + 2 3
dy
= y0 +
y0 +
y0 + .
dx
2
6
dy
= 0. Then
For maxima and minima
dx
y0 +
2u 1 2
3u2 6u + 2 3
y0 +
y0 + = 0
2
6
(7.62)
For simplicity, the third and higher dierences are neglected and obtain the quadratic
equation for u as
au2 + bu + c = 0,
(7.63)
1
1
1
where a = 3 y0 , b = 2 y0 3 y0 , c = y0 2 y0 + 3 y0 .
2
2
3
The values of u can be determined by solving this equation. Then the values of x
are determined from the relation x = x0 + uh. Finally, the maximum value of y can be
obtained from the equation (7.61).
Example 7.9.1 Find x for which y is maximum and also nd the corresponding
value of y, from the following table.
x :
1.5
2.0
2.5
3.0
3.5
y : 0.40547 0.69315 0.91629 1.09861 1.25276
2 y
0.28768
2.0 0.69315
0.06454
0.22314
2.5 0.91629
0.04082
0.18232
3.0 1.09861
0.02817
0.15415
3.5 1.25276
Let x0 = 1.5. Using formula (7.62) we have
2u 1 2
2u 1
y0 = 0 or, 0.28768 +
(0.06454) = 0 or, u = 4.95739.
y0 +
2
2
Therefore, x = x0 + uh = 1.5 + 0.5 4.95739 = 3.97870.
For this x, the value of y is obtained by Newtons backward formula as
y(3.97870) = 1.25276 + 0.47870 (0.15415)
0.47870(0.47870 + 1)
+
(0.02817) = 1.31658.
2
This is the approximate maximum value of y when x = 3.97870.
7.10
Integration
It is well known that, if a function f (x) is known completely, even then it is not always
possible to evaluate the denite integral of it using analytic method. Again, in many
real life problems, we are required to integrate a function between two given limits,
but the function is not known explicitly, but, it is known in a tabular form (equally or
unequally spaced). Then a method, known as numerical integration or quadrature
can be used to solve all such problems.
The problem of numerical integration is stated below:
Given a set of data points (x0 , y0 ), (x1 , y1 ), . . . , (xn , yn ) of a function y = f (x), it is
8b
required to nd the value of the denite integral a f (x) dx. The function f (x) is replaced
by a suitable interpolating polynomial (x).
Then the approximate value of the denite integral is calculated using the following
formula
% b
% b
f (x) dx
(x) dx.
(7.64)
a
433
Thus, dierent integration formulae can be derived depending on the type of the
interpolation formulae used.
A numerical integration formula is said to be of closed type, if the limits of integration a and b are taken as interpolating points. If a and b are not taken as interpolating
points then the formula is known as open type formula.
7.11
(x) = y0 + uy0 +
(7.65)
x x0
, h is the spacing.
h
Let the interval [a, b] be divided into n equal subintervals such that a = x0 < x1 <
x2 < < xn = b. Then
% xn
% b
f (x) dx
(x) dx
I=
where u =
x0
a
xn
=
x0
u2 u 2
u3 3u2 + 2u 3
y0 +
y0 + dx.
y0 + uy0 +
2!
3!
u2 n 2 y u3 u2 n 3 y u4
n
0
0
n
3
2
u +u
= h y0 [u]0 + y0
+
+
+
2 0
2!
3
2 0
3!
4
0
n
2n2 3n 2
n3 4n2 + 4n 3
y0 +
y0 + .
= nh y0 + y0 +
(7.66)
2
12
24
From this formula, one can generate dierent integration formulae by substituting
n = 1, 2, 3, . . . .
7.11.1
Trapezoidal Rule
Substituting n = 1 in the equation (7.66). In this case all dierences higher than the
rst dierence become zero. Then
% xn
h
1
1
f (x) dx = h y0 + y0 = h y0 + (y1 y0 ) = (y0 + y1 ).
(7.67)
2
2
2
x0
x1
xn1
h
h
h
h
[y0 + y1 ] + [y1 + y2 ] + [y2 + y3 ] + + [yn1 + yn ]
2
2
2
2
h
= [y0 + 2(y1 + y2 + + yn1 ) + yn ].
(7.68)
2
Error in trapezoidal rule
The error of trapezoidal rule is
%
E=
a
f (x) dx
h
(y0 + y1 ).
2
(7.69)
Let y = f (x) be continuous and possesses continuous derivatives of all orders. Also,
it is assumed that there exists a function F (x) such that F (x) = f (x) in [x0 , x1 ].
Then
% x1
% b
f (x) dx =
F (x) dx = F (x1 ) F (x0 )
a
x0
h2
F (x0 )
2!
(7.70)
435
Again,
h
h
(y0 + y1 ) = [y0 + y(x0 + h)]
2
2
h
h2
= [y0 + y(x0 ) + hy (x0 ) + y (x0 ) + ]
2
2!
2
h
h
= [y0 + y0 + hy0 + y0 + ].
2
2!
(7.71)
(7.72)
h3
(y + y1 + + yn1
).
12 0
then
If y () is the largest among the n quantities y0 , y1 , . . . , yn1
(b a) 2
1 3
h ny () =
h y (), as nh = b a.
12
12
Note 7.11.1 The error term shows that if the second and higher order derivatives of
f (x) vanish then the trapezoidal rule gives exact result of the integral. This means, the
method gives exact result when f (x) is linear.
Geometrical interpretation of trapezoidal rule
In this rule, the curve y = f (x) is replaced by the line joining the points A(x0 , y0 ) and
B(x1 , y1 ) (Figure 7.1). Thus the area bounded by the curve y = f (x), the ordinates
x = x0 , x = x1 and the xaxis is then approximately equivalent to the area of the
trapezium (ABCD) bounded by the line AB, x = x0 , x = x1 and xaxis.
The geometrical signicance of composite trapezoidal rule is that the curve y = f (x)
is replaced by n straight lines joining the points (x0 , y0 ) and (x1 , y1 ); (x1 , y1 ) and (x2 , y2 );
. . ., (xn1 , yn1 ) and (xn , yn ). Then the area bounded by the curve y = f (x), the lines
x = x0 , x = xn and the xaxis is then approximately equivalent to the sum of the area
of n trapeziums (Figures 7.2).
y = f (x)
B
A
y1
y0
O D x0
C x1
 x
x0
x1
x2
xn
 x
I=
[g(t) + g(t)]dt.
=q
0
g(t) q dt = q
1
%
f (x)dx =
g(t)dt +
g(t)dt
0
437
1
q
t[g (t) g (t)]dt
0
0
% 1
t.2tg (c)dt, where 0 < c < 1
= q[g(1) + g(1)] q
I = q {g(t) + g(t)}t
Program 7.3
.
/* Program Trapezoidal
This program finds the value of integration of a function
by trapezoidal rule.
Here we assume that f(x)=x^3. */
#include<stdio.h>
void main()
{
float a,b,h,sum; int n,i;
float f(float);
printf("Enter the values of a, b ");
scanf("%f %f",&a,&b);
printf("Enter the value of n ");
scanf("%d",&n);
h=(ba)/n;
sum=(f(a)+f(a+n*h))/2.;
for(i=1;i<=n1;i++) sum+=f(a+i*h);
sum=sum*h;
printf("The value of the integration is %8.5f ",sum);
}
/* definition of the function f(x) */
float f(float x)
{
return(x*x*x);
}
A sample of input/output:
Enter the values of a, b 0 1
Enter the value of n 100
The value of the integration is
7.11.2
0.25002
In this formula the interval [a, b] is divided into two equal subintervals by the points
x0 , x1 , x2 , where h = (b a)/2, x1 = x0 + h and x2 = x1 + h.
This rule is obtained by putting n = 2 in (7.66). In this case, the third and higher
order dierences do not exist.
439
% xn
1 2
1
f (x) dx 2h y0 + y0 + y0 = 2h[y0 + (y1 y0 ) + (y2 2y1 + y0 )]
6
6
x0
h
= [y0 + 4y1 + y2 ].
(7.73)
3
The above rule is known as Simpsons 1/3 rule or simply Simpsons rule.
Composite Simpsons 1/3 rule
Let the interval [a, b] be divided into n (an even number) equal subintervals by the
points x0 , x1 , x2 , . . . , xn , where xi = x0 + ih, i = 1, 2, . . . , n. Then
% b
% x2
% x4
% xn
f (x) dx =
f (x) dx +
f (x) dx + +
f (x) dx
x0
x2
xn2
h
h
h
= [y0 + 4y1 + y2 ] + [y2 + 4y3 + y4 ] + + [yn2 + 4yn1 + yn ]
3
3
3
h
= [y0 + 4(y1 + y3 + + yn1 ) + 2(y2 + y4 + + yn2 ) + yn ].
3
(7.74)
This formula is known as Simpsons 1/3 composite rule for numerical integration.
Error in Simpsons 1/3 rule
The error in this formula is
xn
E=
x0
f (x) dx
h
[y0 + 4y1 + y2 ].
3
(7.75)
Let the function f (x) be continuous in [x0 , x2 ] and possesses continuous derivatives
of all order. Also, let there exists a function F (x) in [x0 , x2 ], such that F (x) = f (x).
Then
% x2
% x2
f (x) dx =
F (x) dx = F (x2 ) F (x0 )
x0
x0
(2h)2
F (x0 )
2!
(2h)3
(2h)4 iv
(2h)5 v
F (x0 ) +
F (x0 ) +
F (x0 ) + F (x0 )
3!
4!
5!
4
2
= 2hf (x0 ) + 2h2 f (x0 ) + h3 f (x0 ) + h4 f (x0 )
3
3
4 5 iv
+ h f (x0 ) + .
(7.76)
15
+
(2h)3
(2h)4 iv
+
f (x0 ) +
f (x0 ) +
3!
4!
4
2
= 2hf (x0 ) + 2h2 f (x0 ) + h3 f (x0 ) + h4 f (x0 )
3
3
5
(7.77)
+ h5 f iv (x0 ) + .
18
Using (7.76) and (7.77), equation (7.75) becomes,
E=
4
5 5 iv
h5
h f (x0 ) + f iv (),
15 18
90
(7.78)
E=
441
y
6
parabola
C
y = f (x)
A
B
E
O
x0
x1
D x
x2
x1
0.5
0.75
y1
x2
1.0
1.0
y2
x3
1.5
0.75
y3
x4 x5 x6
2.0 2.5 3.0
0.0 1.25 3.0
y4 y5
y6
h
[y0 + 2(y1 + y2 + y3 + y4 + y5 ) + y6 ]
2
0.5
[0 + 2(0.75 + 1.0 + 0.75 + 0 1.25) 3.0] = 0.125.
=
2
(ii) By Simpsons rule:
83
h
2
0 (2x x ) dx = 3 [y0 + 4(y1 + y3 + y5 ) + 2(y2 + y4 ) + y6 ]
0.5
[0 + 4(0.75 + 0.75 1.25) + 2(1.0 + 0.0) 3.0]
=
3
0.5
[0 + 1 + 2 3] = 0.
=
3
0
(2x x2 ) dx =
f (x)dx = q
I=
%
f (p + qz)dz
1
=q
% 1
% 1
g(z)dz +
g(z)dz = q
[g(z) + g(z)]dz
1
% 0
=q
%
1
1
(z)dz,
=q
(7.80)
To prove
0
c.
%
(z)d(z + c) =
0
1+c %
= y(y c)
1+c
(z)dz =
0
c
1+c
(y c)dy
[where z + c = y]
y (y c)dy
= (1 + c)(1) c(0)
(z + c) (z)d(z + c)
= (1 + c)(1) c(0)
(z + c) (z)dz.
(7.81)
(z + c) (z)dz
0
0
1 % 1 z 2
z 2
+ cz + c1 (z) +
+ cz + c1 (z)dz
= (1 + c)(1) c(0)
2
2
0
0
(z)dz = (1 + c)(1) c(0)
443
1
+ c + c1 (1) + c1 (0)
= (1 + c)(1) c(0)
2
z 3
1 % 1 z 3
z2
z2
+ c + c1 z + c2 (z)
+ c + c1 z + c2 (z)dz
+
6
2
6
2
0
0
1 c
1
+ c + c1 (1) +
+ + c1 + c2 (1)
= (1 + c)(1) c(0)
2
6 2
% 1 3
2
z
z
+ c + c1 z + c2 (z)dz,
c2 (0)
(7.82)
6
2
0
where c1 , c2 , c3 are arbitrary constants and they are chosen in such a way that (1), (1)
and (0) vanish. Thus
1
+ c + c1 = 0,
2
1 c
+ + c1 + c2 = 0,
6 2
and
c2 = 0.
2
1
The solution of these equations is c2 = 0, c1 = , c = .
6
3
Hence
% 1 3
1
2
z 2 z
z
+
(z)dz
I = q (1) + (0)
3
3
6
3
6
0
1
4 a + b h % 1
(z 3 2z 2 + z) (z)dz
=h
f (a) + f (b) + f
3
3
2
6 0
ba
=h
as q =
2
a + b
h
=
f (a) + 4f
+ f (b) + E
3
2
where
%
%
h 1
h 1
2
z(z 1) (z)dz =
z(z 1)2 [g (z) g (z)]dz
E=
6 0
6 0
%
h 1
z(z 1)2 .[2zg iv ()]dz,
z < < z
=
6 0
[by Lagranges MVT]
% 1
h iv
z 2 (z 1)2 dz
[by MVT of integral calculus]
= g (1 )
3
0
h
1
h
0 < 1 < 1.
= g iv (1 ). = g iv (1 ),
3
30
90
%
a
a + b
h
h5
f (a) + 4f
f (x)dx =
+ f (b) f iv (2 ).
3
2
90
Here, the rst term is the value of the integration obtained from the Simpsons 1/3
rule and the second term is its error.
Algorithm 7.4 (Simpsons 1/3).
8b
a f (x) dx using Simpsons 1/3 rule.
445
for(i=1;i<=n1;i+=2)
sum+=4*f(a+i*h)+2*f(a+(i+1)*h);
sum*=h/3.;
printf("Value of the integration is %f ",sum);
} /* main */
/* definition of the function f(x) */
float f(float x)
{
return(x*x*x);
}
A sample of input/output:
Enter the values of a, b 0 1
Enter the value of subintervals n 100
Value of the integration is 0.250000
7.11.3
1 3
y0
8
1
(y3 3y2 + 3y1 y0 )
8
(7.83)
x0
x3
xn3
3h
[(y0 + 3y1 + 3y2 + y3 ) + (y3 + 3y4 + 3y5 + y6 )
=
8
+ + (yn3 + 3yn2 + 3yn1 + yn )]
3h
[y0 + 3(y1 + y2 + y4 + y5 + y7 + y8 + + yn2 + yn1 )
=
8
(7.84)
+2(y3 + y6 + y9 + + yn3 ) + yn ].
Booles rule
% b
5 2
2 3
7 4
f (x)dx = 4h y0 + 2y0 + y0 + y0 + y0
3
3
90
a
5
2
= 4h[y0 + 2(y1 y0 ) + (y2 2y1 + y0 ) + (y3 3y2 + 3y1 y0 )
3
3
7
+ (y4 4y3 + 6y2 4y1 + y0 )]
90
2h
[7y4 + 32y3 + 12y2 + 32y1 + 7y0 ].
(7.85)
=
45
This rule is known as Booles rule.
8h7 vi
f (), a < < b.
It can be shown that the error of this formula is
945
7.11.5
Weddles rule
9 2
41 4
11 5
41 6
3
y0
= 6h y0 + 3y0 + y0 + 4 y0 + y0 + y0 +
2
20
20
840
h 6
9
41
11
1
y0 .
= 6h y0 + 3y0 + 2 y0 +43 y0 + 4 y0 + 5 y0 + 6 y0
2
20
20
20
140
h 6
y0 .
If the sixth order dierence is very small, then we may neglect the last term
140
But, this rejection increases a negligible amount of error, though, it simplies the integration formula. Then the above equation becomes
% x6
f (x)dx
x0
3h
[20y0 + 60y0 + 902 y0 + 803 y0 + 414 y0 + 115 y0 + 6 y0 ]
=
10
3h
[y0 + 5y1 + y2 + 6y3 + y4 + 5y5 + y6 ].
=
10
This formula is known as Weddles rule for numerical integration.
(7.86)
447
x0
x6
xn6
3h
[y0 + 5y1 + y2 + 6y3 + y4 + 5y5 + y6 ]
=
10
3h
+ [y6 + 5y7 + y8 + 6y9 + y10 + 5y11 + y12 ] +
10
3h
+ [yn6 + 5yn5 + yn4 + 6yn3 + yn2 + 5yn1 + yn ]
10
3h
=
[y0 + 5(y1 + y5 + y7 + y11 + + yn5 + yn1 )
10
+(y2 + y4 + y8 + y10 + + yn4 + yn2 )
+6(y3 + y9 + y15 + + yn3 ) + 2(y6 + y12 + + yn6 )].
(7.87)
The above formula is known as Weddles composite rule.
By the technique used in trapezoidal and Simpsons 1/3 rules one can prove that the
h7 vi
f (), x0 < < x6 .
error in Weddles rule is
140
Degree of Precision
The degree of precision of a quadrature formula is a positive integer n such that the
error is zero for all polynomials of degree i n, but it is nonzero for some polynomials
of degree n + 1.
The degree of precision of some quadrature formulae are given in Table 7.2.
Table 7.2: Degree of precision of some quadrature formulae.
Method
Degree of precision
Trapezoidal
1
Simpsons 1/3
3
Simpsons 3/8
3
Booles
5
Weddles
5
7.12
Let the function y = f (x) be known at the (n + 1) points x0 , x1 , . . . , xn of [a, b], these
points need not be equispaced.
The Lagranges interpolation polynomial is
(x) =
n
i=0
w(x)
yi
(x xi )w (xi )
(7.88)
where w(x) = (x x0 ) (x xn )
and (xi ) = yi , i = 0, 1, 2, . . . , n.
If the function f (x) is replaced by the polynomial (x) then
%
f (x)dx
(x)dx =
n %
i=0
w(x)
yi dx.
(x xi )w (xi )
(7.89)
f (x)dx
where Ci =
a
n
Ci yi ,
(7.90)
i=0
b
w(x)
dx,
(x xi )w (xi )
i = 0, 1, 2, . . . , n.
(7.91)
It may be noted that the coecients Ci are independent of the choice of the function
f (x) for a given set of points.
7.13
449
(7.92)
i=0
n
Li (x)yi ,
(7.93)
i=0
where
Li (x) =
(7.94)
Li (x) =
(7.95)
x0
i=0
n
n
(1)ni s(s 1)(s 2) (s n)
yi dx =
or,
Ci yi
(s i)
x0 i=0 i!(n i)!
i=0
n % xn
n
(1)ni s(s1)(s2) (sn)
dx yi =
or,
Ci yi .
(s i)
x0 i!(n i)!
xn
i=0
(7.96)
i=0
(7.97)
n
0
(7.98)
s(s 1)(s 2) (s n)
ds, i = 0, 1, 2, . . . , n.
(s i)
(7.99)
(7.100)
i=0
n
i=0
That is,
% b
n
a i=0
Again,
% b
n
a i=0
w(x)
=1
(x xi )w (xi )
w(x)
dx =
(x xi )w (xi )
w(x)
dx =
(x xi )w (xi )
n
h(1)ni
i=0 0
n
Ci .
dx = (b a).
(7.101)
s(s 1)(s 2) (s n)
ds
i!(n i)!(s i)
(7.102)
i=0
Ci = b a.
(7.103)
(ii)
n
451
Hi = 1.
i=0
n
Ci = (b a)
i=0
n
Hi
i=0
n
or, (b a) = (b a)
Hi . [using (7.103)]
i=0
Hence,
n
Hi = 1.
(7.104)
i=0
(1)i h
(n i)!i!
s(s 1)(s 2) (s n)
ds.
s (n i)
Substituting t = n s, we obtain
Cni
%
(1)i h(1)n 0 t(t 1)(t 2) (t n)
dt
=
i!(n i)!
ti
n
%
(1)ni h n s(s 1)(s 2) (s n)
dt = Ci .
=
i!(n i)! 0
si
Hence,
Ci = Cni .
(7.105)
(iv) Hi = Hni .
Multiplying (7.105) by (b a) and hence obtain
Hi = Hni .
(7.106)
Trapezoidal rule
Substituting n = 1 in (7.100), we get
%
f (x)dx = (b a)
1
Hi yi = (b a)(H0 y0 + H1 y1 ).
i=0
f (x)dx = (b a)
2
Hi yi = (b a)(H0 y0 + H1 y1 + H2 y2 )
i=0
h
(y0 + 4y1 + y2 ).
3
Weddles rule
To deduce the Weddles rule, n = 6 is substituted in (7.100).
%
a
f (x)dx = (b a)
6
H i yi
i=0
= 6h(H0 y0 + H1 y1 + H2 y2 + H3 y3 + H4 y4 + H5 y5 + H6 y6 )
= 6h[H0 (y0 + y6 ) + H1 (y1 + y5 ) + H2 (y2 + y4 ) + H3 y3 ].
To nd the values of Hi s one may use the result Hi = Hni . Also the value of H3
can be obtained by the formula
453
(7.107)
7.14
1
2
1
3
3
8
14
45
95
288
41
140
1
2
4
3
9
8
64
45
375
288
216
140
1
3
9
8
24
45
250
288
27
140
3
8
64
45
250
288
272
140
14
45
375
288
27
140
95
288
216 41
140 140
All the formulae based on NewtonCotes formula developed in Section 7.13 are of closed
type, i.e., they use the function values at the end points a, b of the interval [a, b] of
integration. Here, some formulae are introduced those take the function values at equispaced intermediate points, but, not at the end points. These formulae may be used
x1
1 3
h f (), x0 x1 .
24
(7.108)
x3
f (x)dx =
x0
3h
3h3
[f (x1 ) + f (x2 )] +
f (), x0 x3 .
2
4
(7.109)
x4
f (x)dx =
x0
4h
14h5 iv
[2f (x1 ) f (x2 ) + 2f (x3 )] +
f (),
3
45
x0 x4 .
(7.110)
x5
f (x)dx =
x0
5h
[11f (x1 ) + f (x2 ) + f (x3 ) + 11f (x4 )]
24
+
95h5 iv
f (), x0 x5 .
144
(7.111)
7.15
455
Gaussian Quadrature
(7.112)
i=1
where xi and wi are respectively called nodes and weights and (x) is called the weight
function. Depending on the weight function dierent quadrature formula can be obtained.
The fundamental theorem of Gaussian quadrature states that the optimal nodes of the
mpoint Gaussian quadrature formula are precisely the zeros of the orthogonal polynomial for the same interval and weight function. Gaussian quadrature is optimal because
it ts all polynomial up to degree 2m exactly.
To determine the weights corresponding to the Gaussian nodes xi , compute a Lagranges interpolating polynomial for f (x) by assuming
(x) =
m
(x xj ).
(7.113)
j=1
Then
m
(xj ) =
(xj xi ).
(7.114)
i=1
i=j
m
j=1
(x)
f (xj )
(x xj ) (xj )
(7.115)
for arbitrary points x. Now, determine a set of points xj and wj such that for a weight
function (x) the following relation is valid.
% b
% b
m
(x)(x)
f (xj )dx
(x)(x)dx =
(x xj ) (xj )
a
a
j=1
m
wj f (xj ),
(7.116)
j=1
%
a
(x)(x)
dx.
x xj
(7.117)
(7.118)
Any nite interval [a, b] can be transferred to the interval [1, 1] using linear transformation
x=
b+a
ba
t+
= qt + p.
2
2
(7.119)
Then,
%
f (x) dx =
f (qt + p) q dt.
(7.120)
Thus to study the Gaussian quadrature, we consider the integral in the form
% 1
n
(x)f (x)dx =
wi f (xi ) + E.
(7.121)
1
7.15.1
i=1
(7.122)
i=1
It may be noted that wi and xi are 2n parameters and therefore the weights and
nodes can be determined such that the formula is exact when f (x) is a polynomial of
degree not exceeding 2n 1.
Let
f (x) = c0 + c1 x + c2 x2 + + c2n1 x2n1 .
(7.123)
Therefore,
%
f (x)dx =
1
2
2
= 2c0 + c2 + c4 + .
3
5
When x = xi , equation (7.123) becomes
f (xi ) = c0 + c1 xi + c2 x2i + c3 x3i + + c2n1 x2n1
.
i
(7.124)
457
]
+w2 [c0 + c1 x2 + c2 x22 + + c2n1 x2n1
2
+w3 [c0 + c1 x3 + c2 x23 + + c2n1 x2n1
]
3
+
+wn [c0 + c1 xn + c2 x2n + + c2n1 x2n1
]
n
= c0 (w1 + w2 + + wn ) + c1 (w1 x1 + w2 x2 + + wn xn )
+c2 (w1 x21 + w2 x22 + + wn x2n ) +
+c2n1 (w1 x2n1
+ w2 x2n1
+ + wn x2n1
).
n
1
2
(7.125)
Since (7.124) and (7.125) are identical, compare the coecients of ci , and nd 2n
equations as follows:
w 1 + w2 + + w n = 2
w1 x1 + w2 x2 + + wn xn = 0
w1 x21 + w2 x22 + + wn x2n = 23
2n1
+ w2 x2n1
+ + wn x2n1
= 0.
w1 x1
n
2
(7.126)
f (x)dx =
1
n
wi f (xi ).
(7.127)
i=1
Unfortunately, determination of general solution of the system (7.126) is very complicated. Thus we concentrate for its particular cases.
Case I. When n = 1, the formula is
%
Thus for n = 1,
%
f (x)dx = 2f (0).
1
(7.128)
(7.129)
= 2
= 0
= 23
= 0.
(7.130)
=2
=0
= 23
=0
(f (x) = 1)
(f (x) = x)
(7.131)
(f (x) = x2 )
(f (x) = x3 ).
(7.132)
= 2
= 0
= 23
= 0
= 25
= 0.
459
Гораздо больше, чем просто документы.
Откройте для себя все, что может предложить Scribd, включая книги и аудиокниги от крупных издательств.
Отменить можно в любой момент.